BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//SiPoly//Claim deadlines//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:SiPoly: AI claim deadlines
X-WR-CALDESC:Resolve-by dates of open AI predictions and promises tracked b
 y SiPoly. https://sipoly.si/deadlines/
REFRESH-INTERVAL;VALUE=DURATION:PT12H
X-PUBLISHED-TTL:PT12H
BEGIN:VEVENT
UID:sipoly.si-claim-83
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261029
DTEND;VALUE=DATE:20261030
SUMMARY:Due: Most code for Meta's AI efforts written by AI within 12 to 18 
 months (Mark Zuckerberg)
DESCRIPTION:“I would guess that sometime in the next 12 to 18 months\, we
 'll reach the point where most of the code that's going toward these effor
 ts is written by AI.” (Mark Zuckerberg\, said Apr 29\, 2025)\n\nHow we j
 udge it: Our interpretation: by 29 Oct 2026\, Meta states (or credible rep
 orting shows) that most code for its AI-model efforts is AI-written\, beyo
 nd autocomplete.\n\nScorecard: https://sipoly.si/claim/zuckerberg-most-cod
 e-12-18m/
URL:https://sipoly.si/claim/zuckerberg-most-code-12-18m/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-736
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20261128
DTEND;VALUE=DATE:20261129
SUMMARY:Due: EO 14434: proposed legal definition of "Super Intelligence" du
 e to the President within 60 days (The White House (U.S. executive branch)
 )
DESCRIPTION:“Within 60 days of the date of this order\, the Assistant to 
 the President for Science and Technology (APST)\, in consultation with the
  heads of other agencies as the APST deems appropriate\, shall submit to t
 he President proposed legislative language to establish a Federal definiti
 on of “Super Intelligence” and “SI”” (The White House (U.S. exec
 utive branch)\, said Sep 29\, 2026)\n\nHow we judge it: Our interpretation
 : correct if by 28 Nov 2026 the APST has submitted the proposed legislativ
 e language for a federal definition of "Super Intelligence"\, as shown by 
 a White House or OSTP statement\, a transmitted bill text\, or credible re
 porting\; wrong if credible reporting shows it was not delivered on time.\
 n\nScorecard: https://sipoly.si/claim/eo14434-si-definition-60d/
URL:https://sipoly.si/claim/eo14434-si-definition-60d/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-106
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261130
DTEND;VALUE=DATE:20261201
SUMMARY:Due: AI will be an election issue in the 2026 US midterms (Gary Mar
 cus)
DESCRIPTION:“In the midterms\, AI will be an election issue for first tim
 e.” (Gary Marcus\, said Dec 20\, 2025)\n\nHow we judge it: Our interpret
 ation: AI appears among top issues in national exit polls or major-party c
 ampaign platforms for Nov 2026.\n\nScorecard: https://sipoly.si/claim/marc
 us-ai-midterms-issue/
URL:https://sipoly.si/claim/marcus-ai-midterms-issue/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-836
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: By end of 2026 OpenAI will have an internal system Altman woul
 d call AGI (Sam Altman)
DESCRIPTION:“by the end of the year the company would have an internal sy
 stem he would call AGI” (Sam Altman\, said Aug 26\, 2026)\n\nHow we judg
 e it: Our interpretation: by 31 Dec 2026\, OpenAI publicly states it has (
 or demonstrably deploys internally) a system meeting OpenAI's charter AGI 
 definition—highly autonomous systems that outperform humans at most econ
 omically valuable work—not merely a self-label without that substance. S
 elf-declaration alone without charter-aligned evidence → needs editor.\n
 \nScorecard: https://sipoly.si/claim/altman-agi-end-2026/
URL:https://sipoly.si/claim/altman-agi-end-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-373
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: First clinical trials of its AI-designed drugs by end of 2026 
 (Isomorphic Labs)
DESCRIPTION:“Isomorphic Labs\, which uses artificial intelligence for dru
 g discovery\, expects to have its first clinical trials by the end of 2026
 \, founder and CEO Demis Hassabis said on Tuesday.” (Isomorphic Labs\, s
 aid Jan 20\, 2026)\n\nHow we judge it: Our interpretation: Isomorphic Labs
  starts a human clinical trial of one of its drug candidates by 31 Dec 202
 6.\n\nScorecard: https://sipoly.si/claim/isomorphic-trials-end-2026/
URL:https://sipoly.si/claim/isomorphic-trials-end-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-698
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: "Probably" by Dec 2026 the recursive self-improvement loop on 
 algorithms will be closed (David "davidad" Dalrymple)
DESCRIPTION:“I would guess that by December 2026 the RSI loop on algorith
 ms will probably be closed\, resulting in another inflection point to an e
 ven faster pace\, perhaps around 70-80 day doubling time.” (David "david
 ad" Dalrymple\, said Dec 20\, 2025)\n\nHow we judge it: Our interpretation
 : correct if by 31 Dec 2026 (1) METR or an equivalent tracker reports that
  the doubling time of frontier 50% time horizons over 2026 is about 80 day
 s or less\, and (2) at least one frontier lab states publicly that AI syst
 ems now do most of its algorithmic research work. Wrong if neither holds\;
  partly if only one does.\n\nScorecard: https://sipoly.si/claim/davidad-rs
 i-loop-dec-2026/
URL:https://sipoly.si/claim/davidad-rsi-loop-dec-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-130
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: 2025 will be seen as the year of "peak bubble" for generative 
 AI (Gary Marcus)
DESCRIPTION:“2025 will be known as the year of the peak bubble \, and als
 o the moment at which Wall Street began to lose confidence in generative A
 I.” (Gary Marcus\, said Dec 20\, 2025)\n\nHow we judge it: Our interpret
 ation: by end-2026\, major AI-exposed equity indices or AI company valuati
 ons are materially below their 2025 peaks.\n\nScorecard: https://sipoly.si
 /claim/marcus-peak-bubble/
URL:https://sipoly.si/claim/marcus-peak-bubble/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-107
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: No country takes a decisive lead in the GenAI race in 2026 (Ga
 ry Marcus)
DESCRIPTION:“No country will take a decisive lead in the GenAI “race”
  .” (Gary Marcus\, said Dec 20\, 2025)\n\nHow we judge it: "Decisive lea
 d" undefined\; our interpretation: frontier-model leaderboards at end-2026
  still show top models from more than one country within a small margin.\n
 \nScorecard: https://sipoly.si/claim/marcus-no-decisive-lead/
URL:https://sipoly.si/claim/marcus-no-decisive-lead/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-105
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: Home humanoid robots "all demo and very little product" in 202
 6 (Gary Marcus)
DESCRIPTION:“Human domestic robots like Optimus and Figure will be all de
 mo and very little product.” (Gary Marcus\, said Dec 20\, 2025)\n\nHow w
 e judge it: Our interpretation: fewer than ~10\,000 general-purpose humano
 id robots delivered to private homes worldwide in 2026.\n\nScorecard: http
 s://sipoly.si/claim/marcus-domestic-robots-2026/
URL:https://sipoly.si/claim/marcus-domestic-robots-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-347
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: In 2026 AI will be able to make "very small discoveries" (Open
 AI)
DESCRIPTION:“In 2026\, we expect AI to be capable of making very small di
 scoveries.” (OpenAI\, said Nov 6\, 2025)\n\nHow we judge it: Our interpr
 etation: during 2026\, AI systems are credited (by the discoverers\, in pu
 blic write-ups) with at least one small but genuinely new scientific or ma
 thematical result.\n\nScorecard: https://sipoly.si/claim/openai-small-disc
 overies-2026/
URL:https://sipoly.si/claim/openai-small-discoveries-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-691
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: 10% (and rising) chance that Grok 5 achieves AGI (Elon Musk)
DESCRIPTION:“My estimate of the probability of Grok 5 achieving AGI is no
 w at 10% and rising” (Elon Musk\, said Oct 18\, 2025)\n\nHow we judge it
 : Our interpretation: YES if Grok 5\, once released\, is broadly recognize
 d as AGI by independent experts (not just by xAI). Scored at the end of 20
 26\, or once Grok 5 has been out for 6 months if later.\n\nScorecard: http
 s://sipoly.si/claim/musk-grok5-agi-10pct/
URL:https://sipoly.si/claim/musk-grok5-agi-10pct/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-382
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: Fully driverless Waymo rides in London in 2026 (Waymo)
DESCRIPTION:“We’re bringing our fully autonomous ride-hailing service a
 cross the pond\, where we intend to offer rides – with no human behind t
 he wheel – in 2026.” (Waymo\, said Oct 2025)\n\nHow we judge it: Our i
 nterpretation: members of the public can take rides with no human behind t
 he wheel in London by 31 Dec 2026.\n\nScorecard: https://sipoly.si/claim/w
 aymo-london-2026/
URL:https://sipoly.si/claim/waymo-london-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-357
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: At least one model will match human experts across many indust
 ries before end-2026 (Julian Schrittwieser)
DESCRIPTION:“At least one model will match the performance of human exper
 ts across many industries before the end of 2026.” (Julian Schrittwieser
 \, said Sep 27\, 2025)\n\nHow we judge it: Our interpretation: by 31 Dec 2
 026\, a model wins or ties at least 50% of comparisons against industry ex
 perts on GDPval (or an equivalent cross-industry evaluation).\n\nScorecard
 : https://sipoly.si/claim/schrittwieser-expert-parity-2026/
URL:https://sipoly.si/claim/schrittwieser-expert-parity-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-342
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: Owners able to add their cars to the Tesla robotaxi fleet in 2
 026\, "confidently next year" (Tesla)
DESCRIPTION:“But I guess next year is I'd say confidently next year. I'm 
 not sure when next year\, but confidently next year\, people would be able
  to add or subtract their car to the Tesla\, Inc. fleet.” (Tesla\, said 
 Jul 23\, 2025)\n\nHow we judge it: Our interpretation: during 2026\, Tesla
  lets private owners put their own cars into its paid robotaxi network.\n\
 nScorecard: https://sipoly.si/claim/tesla-add-own-car-to-fleet-2026/
URL:https://sipoly.si/claim/tesla-add-own-car-to-fleet-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-693
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: Prometheus\, a 1 GW AI supercluster\, online in 2026 (Meta)
DESCRIPTION:“We're actually building several multi-GW clusters. We're cal
 ling the first one Prometheus and it's coming online in '26.” (Meta\, sa
 id Jul 14\, 2025)\n\nHow we judge it: Our interpretation: correct if durin
 g 2026 Meta's Prometheus cluster (New Albany\, Ohio) is operating at about
  1 GW or more\, per Meta or Epoch AI\; partly if it comes online in 2026 b
 ut well below 1 GW\; wrong if it is not online in 2026.\n\nScorecard: http
 s://sipoly.si/claim/meta-prometheus-1gw-2026/
URL:https://sipoly.si/claim/meta-prometheus-1gw-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-63
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: 2026 will likely see AI systems that figure out novel insights
  (Sam Altman)
DESCRIPTION:“2026 will likely see the arrival of systems that can figure 
 out novel insights.” (Sam Altman\, said Jun 10\, 2025)\n\nHow we judge i
 t: Our interpretation: during 2026 an AI system is credited\, in a peer-re
 viewed paper or by recognised domain experts\, as the principal originator
  of a genuinely new scientific or mathematical result.\n\nScorecard: https
 ://sipoly.si/claim/altman-2026-novel-insights/
URL:https://sipoly.si/claim/altman-2026-novel-insights/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-717
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: AI better\, faster and cheaper at any screen-based job "by nex
 t year" (Emad Mostaque)
DESCRIPTION:“For any job that you can do on the other side of a screen\, 
 an AI will probably be able to do it better\, faster\, and cheaper by next
  year.” (Emad Mostaque\, said Jun 2025)\n\nHow we judge it: Our interpre
 tation: correct if by 31 Dec 2026 AI can do essentially any remote (screen
 -based) job better\, faster and more cheaply than the humans doing it.\n\n
 Scorecard: https://sipoly.si/claim/mostaque-screen-jobs-by-2026/
URL:https://sipoly.si/claim/mostaque-screen-jobs-by-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-716
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: Digital superintelligence "this year" (2025)\, or "next year f
 or sure" (Elon Musk)
DESCRIPTION:“I think we're quite close to digital superintelligence. It m
 ay happen this year and if it doesn't happen this year\, next year for sur
 e. A digital superintelligence defined as smarter than any human at anythi
 ng.” (Elon Musk\, said Jun 2025)\n\nHow we judge it: Our interpretation:
  correct if by 31 Dec 2026 an AI system is broadly acknowledged as smarter
  than any human at anything (superhuman in essentially every cognitive dom
 ain).\n\nScorecard: https://sipoly.si/claim/musk-digital-superintelligence
 -2026/
URL:https://sipoly.si/claim/musk-digital-superintelligence-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-385
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: First 200 MW of Stargate UAE expected to go live in 2026 (Open
 AI)
DESCRIPTION:“A 1GW Stargate UAE cluster in Abu Dhabi with 200MW expected 
 to go live in 2026” (OpenAI\, said May 22\, 2025)\n\nHow we judge it: Ou
 r interpretation: the first 200 MW of Stargate UAE is operational by 31 De
 c 2026.\n\nScorecard: https://sipoly.si/claim/openai-stargate-uae-200mw-20
 26/
URL:https://sipoly.si/claim/openai-stargate-uae-200mw-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-445
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: A 10\,000-person anti-AI protest in Washington\, DC by late 20
 26 (AI Futures Project)
DESCRIPTION:“Many people fear that the next wave of AIs will come for the
 ir jobs\; there is a 10\,000 person anti-AI protest in DC.” (AI Futures 
 Project\, said Apr 3\, 2025)\n\nHow we judge it: Our interpretation: corre
 ct if a protest against AI with roughly 10\,000 or more participants (cred
 ible media estimates) takes place in Washington\, DC by 31 Dec 2026.\n\nSc
 orecard: https://sipoly.si/claim/ai2027-10k-anti-ai-protest-dc-2026/
URL:https://sipoly.si/claim/ai2027-10k-anti-ai-protest-dc-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-444
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: The stock market will be up 30% in 2026\, led by AI companies 
 (AI Futures Project)
DESCRIPTION:“The stock market has gone up 30% in 2026\, led by OpenBrain\
 , Nvidia\, and whichever companies have most successfully integrated AI as
 sistants.” (AI Futures Project\, said Apr 3\, 2025)\n\nHow we judge it: 
 Our interpretation: correct if the S&P 500 total return for calendar 2026 
 is at least +25%\, with AI-linked companies leading.\n\nScorecard: https:/
 /sipoly.si/claim/ai2027-stock-market-up-30pct-2026/
URL:https://sipoly.si/claim/ai2027-stock-market-up-30pct-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-381
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: Waymo One ready for riders in Washington\, D.C. in 2026 (Waymo
 )
DESCRIPTION:“It’s official: Waymo One\, the world’s leading fully aut
 onomous ride-hailing service\, will be ready for riders in the nation’s 
 capital on the Waymo One app in 2026.” (Waymo\, said Mar 2025)\n\nHow we
  judge it: Our interpretation: public riders can take fully autonomous Way
 mo rides in Washington\, D.C. by 31 Dec 2026.\n\nScorecard: https://sipoly
 .si/claim/waymo-dc-2026/
URL:https://sipoly.si/claim/waymo-dc-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-384
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: Vera Rubin GPUs to ship in the second half of 2026 (Nvidia)
DESCRIPTION:“The centerpiece announcement was Vera Rubin\, first teased a
 t Computex 2024 and now scheduled for release in the second half of 2026.
 ” (Nvidia\, said Mar 18\, 2025)\n\nHow we judge it: Our interpretation: 
 Vera Rubin systems are commercially available from Nvidia partners by 31 D
 ec 2026.\n\nScorecard: https://sipoly.si/claim/nvidia-vera-rubin-h2-2026/
URL:https://sipoly.si/claim/nvidia-vera-rubin-h2-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-324
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: 40% chance of AGI (replacing the majority of human jobs) by en
 d of 2026 (Andrew Critch)
DESCRIPTION:“p(AGI by eoy 2026) = 40%” (Andrew Critch\, said Feb 6\, 20
 25)\n\nHow we judge it: Our interpretation: AI that\, at runtime and for l
 ess than a human costs\, can replace humans in the power-weighted majority
  of jobs exists by 31 Dec 2026.\n\nScorecard: https://sipoly.si/claim/crit
 ch-agi-2026-40pct/
URL:https://sipoly.si/claim/critch-agi-2026-40pct/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-127
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: Optimus "production design 2" to launch in 2026 (Tesla)
DESCRIPTION:“Those Optimus in use at the Tesla factories for production d
 esign 1 will inform how will we change for production design 2\, which we 
 expect to launch next year.” (Tesla\, said Jan 29\, 2025)\n\nHow we judg
 e it: Tesla starts production of a second Optimus production design during
  2026.\n\nScorecard: https://sipoly.si/claim/tesla-optimus-v2-2026/
URL:https://sipoly.si/claim/tesla-optimus-v2-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-74
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: Powerful AI "could come as early as 2026" (Dario Amodei)
DESCRIPTION:“I think it could come as early as 2026\, though there are al
 so ways it could take much longer.” (Dario Amodei\, said Oct 2024)\n\nHo
 w we judge it: Our interpretation: a system meeting the essay's "powerful 
 AI" definition exists by 31 Dec 2026.\n\nScorecard: https://sipoly.si/clai
 m/amodei-powerful-ai-2026/
URL:https://sipoly.si/claim/amodei-powerful-ai-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-118
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: A new chip shortage in 2026 (Daniel Kokotajlo)
DESCRIPTION:“We’re in a new chip shortage. Just when the fabs thought t
 hey had caught up to demand…” (Daniel Kokotajlo\, said Aug 6\, 2021)\n
 \nHow we judge it: Our interpretation: during 2026\, widely reported AI-dr
 iven shortages of leading-edge chips or memory.\n\nScorecard: https://sipo
 ly.si/claim/kokotajlo-chip-shortage-2026/
URL:https://sipoly.si/claim/kokotajlo-chip-shortage-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-435
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20261231
DTEND;VALUE=DATE:20270101
SUMMARY:Due: AI will outperform humans at writing high-school essays by 202
 6 (median) (ML researchers (2016 expert survey))
DESCRIPTION:“Researchers predict AI will outperform humans in many activi
 ties in the next ten years\, such as translating languages (by 2024)\, wri
 ting high-school essays (by 2026)\, driving a truck (by 2027)\, working in
  retail (by 2031)\, writing a bestselling book (by 2049)\, and working as 
 a surgeon (by 2053).” (ML researchers (2016 expert survey)\, said May 24
 \, 2017)\n\nHow we judge it: Our interpretation: by end-2026\, AI-written 
 high-school essays reliably score at or above typical human students in bl
 ind grading.\n\nScorecard: https://sipoly.si/claim/ml-survey-2016-hs-essay
 s-2026/
URL:https://sipoly.si/claim/ml-survey-2016-hs-essays-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-723
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20270219
DTEND;VALUE=DATE:20270220
SUMMARY:Due: About a 50% chance of AGI within 3 years (Feb 2024) (Daniel Ko
 kotajlo)
DESCRIPTION:“I expect to need the money sometime in the next 3 years\, be
 cause that’s about when we get to 50% chance of AGI.” (Daniel Kokotajl
 o\, said Feb 19\, 2024)\n\nHow we judge it: Our interpretation: probabilis
 tic (50%). YES if by 19 Feb 2027 an AI system is broadly recognized by ind
 ependent experts as AGI (able to do essentially all the cognitive work of 
 a top AI researcher/engineer)\; a developer's own claim does not count on 
 its own.\n\nScorecard: https://sipoly.si/claim/kokotajlo-50pct-agi-by-2027
 /
URL:https://sipoly.si/claim/kokotajlo-50pct-agi-by-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-327
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20270312
DTEND;VALUE=DATE:20270313
SUMMARY:Due: "At least 3-5 years away from automating software engineering"
  (from Mar 2024) (Bindu Reddy)
DESCRIPTION:“We are at least 3-5 years away from automating software engi
 neering.” (Bindu Reddy\, said Mar 12\, 2024)\n\nHow we judge it: Our int
 erpretation: correct if software engineering is not automated (AI doing th
 e end-to-end work of most professional software engineers) before 12 Mar 2
 027.\n\nScorecard: https://sipoly.si/claim/reddy-swe-automation-3-5y/
URL:https://sipoly.si/claim/reddy-swe-automation-3-5y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-730
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20270331
DTEND;VALUE=DATE:20270401
SUMMARY:Due: SpaceX's AI effort to reach "pole position" in about six month
 s (from Sep 2026) (Elon Musk)
DESCRIPTION:“If our second derivative remains strong\, SpaceX will reach 
 pole position in about 6 months.” (Elon Musk\, said Sep 24\, 2026)\n\nHo
 w we judge it: Our interpretation: correct if by 31 Mar 2027 a SpaceX/xAI 
 model ranks #1 overall on the Artificial Analysis Intelligence Index or ho
 lds the top overall text score on LMArena (leading the frontier overall\, 
 not just a sub-index).\n\nScorecard: https://sipoly.si/claim/musk-spacex-a
 i-pole-position-6mo/
URL:https://sipoly.si/claim/musk-spacex-ai-pole-position-6mo/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-732
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20270331
DTEND;VALUE=DATE:20270401
SUMMARY:Due: Embedded third-party evaluators with "employee-like access" in
 side Anthropic (Anthropic)
DESCRIPTION:“First\, we committed to embed external evaluators inside Ant
 hropic with employee-like access\, similar to a food inspector\, and we re
 commended that other companies across the world do the same.” (Anthropic
 \, said Sep 23\, 2026)\n\nHow we judge it: Our interpretation: correct if 
 by 31 Mar 2027 Anthropic and at least one named independent evaluator (e.g
 . METR) publicly confirm that an embedded third-party evaluation team has 
 ongoing\, employee-like access inside Anthropic (to models and training pi
 pelines)\; wrong if no such arrangement is confirmed by then.\n\nScorecard
 : https://sipoly.si/claim/anthropic-embedded-evaluators-2026/
URL:https://sipoly.si/claim/anthropic-embedded-evaluators-2026/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-72
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20270331
DTEND;VALUE=DATE:20270401
SUMMARY:Due: Powerful AI systems in late 2026 or early 2027 (Anthropic)
DESCRIPTION:“we expect powerful AI systems will emerge in late 2026 or ea
 rly 2027.” (Anthropic\, said Mar 6\, 2025)\n\nHow we judge it: Our inter
 pretation\, using Anthropic's own definition: by 31 Mar 2027 a system exis
 ts matching or exceeding Nobel-level researchers across most listed discip
 lines.\n\nScorecard: https://sipoly.si/claim/anthropic-powerful-ai-2026-27
 /
URL:https://sipoly.si/claim/anthropic-powerful-ai-2026-27/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-838
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20270605
DTEND;VALUE=DATE:20270606
SUMMARY:Due: Claude-written code expected to be strictly better than human-
 written code within the year (Anthropic)
DESCRIPTION:“Claude-written code was somewhat worse than human-written co
 de at Anthropic in late 2025\, is roughly at parity today\, and we expect 
 it to be strictly better within the year.” (Anthropic\, said Jun 5\, 202
 6)\n\nHow we judge it: Our interpretation: by 5 Jun 2027 (12 months from d
 ate_said)\, Anthropic or a credible independent evaluator reports that Cla
 ude-authored production code at Anthropic (or equivalent) is strictly bett
 er than typical human-authored code on Anthropic's own stated quality crit
 eria (correctness + maintainability).\n\nScorecard: https://sipoly.si/clai
 m/anthropic-claude-code-better-within-year/
URL:https://sipoly.si/claim/anthropic-claude-code-better-within-year/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-731
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20270630
DTEND;VALUE=DATE:20270701
SUMMARY:Due: OpenAI's in-house chip program to start coming online in the f
 irst half of 2027 (OpenAI)
DESCRIPTION:“which will start to come online in the first half of next of
  2027 and will scale a lot in future years” (OpenAI\, said Sep 29\, 2026
 )\n\nHow we judge it: Our interpretation: correct if by 30 Jun 2027 OpenAI
 's own custom AI chips are running production workloads (e.g. serving Chat
 GPT or API inference)\, as confirmed by OpenAI or reputable reporting. Tap
 e-outs or test deployments alone do not count.\n\nScorecard: https://sipol
 y.si/claim/openai-custom-chips-h1-2027/
URL:https://sipoly.si/claim/openai-custom-chips-h1-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-734
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20270630
DTEND;VALUE=DATE:20270701
SUMMARY:Due: Cybercab operating commercially in California "probably by the
  middle of next year" (2027) (Tesla)
DESCRIPTION:“And\, probably\, by the middle of next year\, it will be ope
 rating commercially in California.” (Tesla\, said Sep 26\, 2026)\n\nHow 
 we judge it: Our interpretation: correct if by 30 Jun 2027 Tesla's Cyberca
 b (the two-seat vehicle with no steering wheel or pedals) carries paying m
 embers of the public in commercial service somewhere in California.\n\nSco
 recard: https://sipoly.si/claim/tesla-cybercab-california-mid-2027/
URL:https://sipoly.si/claim/tesla-cybercab-california-mid-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-718
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20270717
DTEND;VALUE=DATE:20270718
SUMMARY:Due: AI "world-class mathematicians" within a year\, world-class pr
 ogrammers within one or two (Jul 2025) (Eric Schmidt)
DESCRIPTION:“So\, it's likely in my opinion that you're going to see worl
 d-class mathematicians emerge in the next one year that are AI based and w
 orld-class programmers that going to appear within the next one or two yea
 rs.” (Eric Schmidt\, said Jul 17\, 2025)\n\nHow we judge it: Our interpr
 etation: correct if by 17 Jul 2026 an AI system is recognized by leading m
 athematicians as doing world-class research mathematics (e.g. original res
 ults accepted in top journals) AND by 17 Jul 2027 AI systems perform at wo
 rld-class level in programming competitions and real-world engineering\; p
 artly if only one half holds.\n\nScorecard: https://sipoly.si/claim/schmid
 t-ai-world-class-mathematicians-1y/
URL:https://sipoly.si/claim/schmidt-ai-world-class-mathematicians-1y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-389
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20270812
DTEND;VALUE=DATE:20270813
SUMMARY:Due: Most white-collar tasks automated within 12 to 18 months (Must
 afa Suleyman)
DESCRIPTION:“said in an interview with the Financial Times that he predic
 ts most\, if not every\, task in white-collar fields will be automated by 
 AI within the next year or year and a half.” (Mustafa Suleyman\, said Fe
 b 12\, 2026)\n\nHow we judge it: Our interpretation: by 12 Aug 2027\, AI c
 an fully perform most tasks of computer-based professional jobs (lawyers\,
  accountants\, project managers\, marketers)\, as shown by broad deploymen
 t or independent evaluations.\n\nScorecard: https://sipoly.si/claim/suleym
 an-white-collar-automated-18m/
URL:https://sipoly.si/claim/suleyman-white-collar-automated-18m/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-841
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20271008
DTEND;VALUE=DATE:20271009
SUMMARY:Due: An AI-led cyberattack steals complete weights of a closed fron
 tier model (within ~12 months) (Nathan Benaich)
DESCRIPTION:“An AI-led cyberattack steals the complete weights of a close
 d frontier model from a leading AI lab.” (Nathan Benaich\, said Oct 8\, 
 2026)\n\nHow we judge it: Our interpretation: YES if by 8 Oct 2027 a credi
 ble report (lab disclosure\, government\, or major outlet with primary sou
 rcing) confirms that an AI agent/system led a cyberattack that exfiltrated
  the complete model weights of a closed frontier model from a leading AI l
 ab.\n\nScorecard: https://sipoly.si/claim/stateofai-frontier-weights-stole
 n-2027/
URL:https://sipoly.si/claim/stateofai-frontier-weights-stolen-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-840
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20271008
DTEND;VALUE=DATE:20271009
SUMMARY:Due: An autonomous AI team beats human-led model research on equal 
 time and compute\, setting its agenda (within ~12 months) (Nathan Benaich)
DESCRIPTION:“An autonomous AI team beats human-led model research on equa
 l time and compute\, setting its agenda.” (Nathan Benaich\, said Oct 8\,
  2026)\n\nHow we judge it: Our interpretation: YES if by 8 Oct 2027 a cred
 ible public report shows an AI system/team that (1) sets its own model-res
 earch agenda and (2) outperforms a human-led team given equal time and com
 pute on that research.\n\nScorecard: https://sipoly.si/claim/stateofai-aut
 onomous-research-team-2027/
URL:https://sipoly.si/claim/stateofai-autonomous-research-team-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-721
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20271024
DTEND;VALUE=DATE:20271025
SUMMARY:Due: "Only two years away from the first AI feature" (Oct 2025) (Pa
 ul Schrader)
DESCRIPTION:“I think we’re only two years away from the first AI featur
 e.” (Paul Schrader\, said Oct 24\, 2025)\n\nHow we judge it: Our interpr
 etation: correct if by 24 Oct 2027 a feature-length film (at least 75 minu
 tes) made predominantly with generative AI gets a commercial theatrical or
  major-streamer release.\n\nScorecard: https://sipoly.si/claim/schrader-fi
 rst-ai-feature-2y/
URL:https://sipoly.si/claim/schrader-first-ai-feature-2y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-839
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: AGI in 2027 (State of AI Report ninth prediction) (Nathan Bena
 ich)
DESCRIPTION:“AI progress is moving so quickly that we expect AGI in 2027\
 , the tenth anniversary of the State of AI Report. Our ninth prediction\, 
 “AGI 2027\,” reflects that conviction about the pace of progress.” (
 Nathan Benaich\, said Oct 8\, 2026)\n\nHow we judge it: Our interpretation
 : YES if by 31 Dec 2027 a broadly recognized AGI milestone is met (multipl
 e major labs or a major scientific/policy body concur that AGI has been ac
 hieved)\, not merely one lab's self-declaration under a private definition
 .\n\nScorecard: https://sipoly.si/claim/stateofai-agi-2027/
URL:https://sipoly.si/claim/stateofai-agi-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-837
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: In 2027\, AI systems could complete tasks that take a person w
 eeks (Anthropic)
DESCRIPTION:“If this trend holds\, tasks that take a skilled person days 
 could come into range this year. In 2027\, AI systems could be capable of 
 tasks that take a person weeks.” (Anthropic\, said Jun 5\, 2026)\n\nHow 
 we judge it: Our interpretation: YES if by 31 Dec 2027\, a credible indepe
 ndent evaluator (e.g. METR) or frontier-lab public eval shows AI agents re
 liably completing software/research tasks that take skilled humans on the 
 order of one or more weeks (≈40+ hours) at the lab's stated reliability 
 threshold (e.g. 50% time horizon).\n\nScorecard: https://sipoly.si/claim/a
 nthropic-week-tasks-2027/
URL:https://sipoly.si/claim/anthropic-week-tasks-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-104
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: No AGI in 2026 (or 2027) (Gary Marcus)
DESCRIPTION:“We won’t get to AGI in 2026 (or 7).” (Gary Marcus\, said
  Dec 20\, 2025)\n\nHow we judge it: No system broadly accepted as AGI by 3
 1 Dec 2027.\n\nScorecard: https://sipoly.si/claim/marcus-no-agi-2026-27/
URL:https://sipoly.si/claim/marcus-no-agi-2026-27/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-318
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: "Something pretty clearly superhuman in most respects by end o
 f 2027" (Miles Brundage)
DESCRIPTION:“very roughly\, something pretty clearly superhuman in most r
 espects by end of 2027 + also very big stuff before then” (Miles Brundag
 e\, said Nov 23\, 2025)\n\nHow we judge it: Our interpretation: by 31 Dec 
 2027\, an AI system outperforms top human experts in most cognitive domain
 s\, per broad expert consensus.\n\nScorecard: https://sipoly.si/claim/brun
 dage-superhuman-by-2027/
URL:https://sipoly.si/claim/brundage-superhuman-by-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-412
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: 10%: AI solves or substantially helps solve a Millennium Prize
  Problem by 2027 (LEAP expert panel)
DESCRIPTION:“Experts estimate a 10% chance that AI will solve or substant
 ially assist in solving a Millennium Prize Problem by 2027” (LEAP expert
  panel\, said Nov 10\, 2025)\n\nHow we judge it: Our interpretation: YES i
 f by 31 Dec 2027 a Millennium Prize Problem is accepted as solved\, with A
 I credited as solving it or substantially assisting.\n\nScorecard: https:/
 /sipoly.si/claim/leap-millennium-prize-2027-10pct/
URL:https://sipoly.si/claim/leap-millennium-prize-2027-10pct/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-358
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: By end-2027\, models will frequently outperform experts on man
 y tasks (Julian Schrittwieser)
DESCRIPTION:“By the end of 2027\, models will frequently outperform exper
 ts on many tasks.” (Julian Schrittwieser\, said Sep 27\, 2025)\n\nHow we
  judge it: Our interpretation: by 31 Dec 2027\, a model clearly beats (not
  just ties) industry experts on a majority of tasks in GDPval or an equiva
 lent evaluation.\n\nScorecard: https://sipoly.si/claim/schrittwieser-outpe
 rform-experts-2027/
URL:https://sipoly.si/claim/schrittwieser-outperform-experts-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-370
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: AGI "likely to arrive by 2027" (Roman Yampolskiy)
DESCRIPTION:“said artificial general intelligence — systems as capable 
 as humans across domains — is likely to arrive by 2027.” (Roman Yampol
 skiy\, said Sep 4\, 2025)\n\nHow we judge it: Our interpretation: AI as ca
 pable as humans across domains exists by 31 Dec 2027\, by broad expert rec
 ognition.\n\nScorecard: https://sipoly.si/claim/yampolskiy-agi-2027/
URL:https://sipoly.si/claim/yampolskiy-agi-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-337
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: Over 40% of agentic AI projects canceled by end of 2027 (Gartn
 er)
DESCRIPTION:“Over 40% of agentic AI projects will be canceled by the end 
 of 2027\, due to escalating costs\, unclear business value or inadequate r
 isk controls” (Gartner\, said Jun 25\, 2025)\n\nHow we judge it: Our int
 erpretation: Gartner or other credible surveys covering 2027 report that m
 ore than 40% of enterprise agentic-AI projects were canceled.\n\nScorecard
 : https://sipoly.si/claim/gartner-agentic-40pct-canceled-2027/
URL:https://sipoly.si/claim/gartner-agentic-40pct-canceled-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-64
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: 2027 may see robots that can do tasks in the real world (Sam A
 ltman)
DESCRIPTION:“2027 may see the arrival of robots that can do tasks in the 
 real world.” (Sam Altman\, said Jun 10\, 2025)\n\nHow we judge it: Our i
 nterpretation: by end of 2027\, general-purpose robots are commercially de
 ployed doing varied real-world tasks outside controlled demos.\n\nScorecar
 d: https://sipoly.si/claim/altman-2027-robots/
URL:https://sipoly.si/claim/altman-2027-robots/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-430
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: Rubin Ultra (NVL576) in the second half of 2027 (Nvidia)
DESCRIPTION:“Huang also announced Rubin Ultra\, which will follow in the 
 second half of 2027.” (Nvidia\, said Mar 18\, 2025)\n\nHow we judge it: 
 Our interpretation: Nvidia ships Rubin Ultra (NVL576 systems) to customers
  by 31 Dec 2027.\n\nScorecard: https://sipoly.si/claim/nvidia-rubin-ultra-
 h2-2027/
URL:https://sipoly.si/claim/nvidia-rubin-ultra-h2-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-719
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: An AI company will claim AGI "probably in 2026 or 2027" (Kevin
  Roose)
DESCRIPTION:“I believe that very soon — probably in 2026 or 2027\, but 
 possibly as soon as this year — one or more A.I. companies will claim th
 ey’ve created an artificial general intelligence” (Kevin Roose\, said 
 Mar 14\, 2025)\n\nHow we judge it: Our interpretation: correct if by 31 De
 c 2027 a major AI company formally and publicly claims to have created AGI
 .\n\nScorecard: https://sipoly.si/claim/roose-lab-claims-agi-2026-27/
URL:https://sipoly.si/claim/roose-lab-claims-agi-2026-27/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-364
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: "Strikingly plausible" that models do the work of an AI resear
 cher/engineer by 2027 (Leopold Aschenbrenner)
DESCRIPTION:“I make the following claim: it is strikingly plausible that 
 by 2027\, models will be able to do the work of an AI researcher/engineer.
 ” (Leopold Aschenbrenner\, said Jun 4\, 2024)\n\nHow we judge it: Our in
 terpretation: by 31 Dec 2027\, AI systems can do the job of a frontier-lab
  AI researcher/engineer (not just assist)\, per lab statements and indepen
 dent evaluations.\n\nScorecard: https://sipoly.si/claim/aschenbrenner-ai-r
 esearcher-2027/
URL:https://sipoly.si/claim/aschenbrenner-ai-researcher-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-440
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: 10% chance machines outperform humans at every task by 2027 (A
 I researchers (2023 expert survey))
DESCRIPTION:“If science continues undisrupted\, the chance of unaided mac
 hines outperforming humans in every possible task was estimated at 10% by 
 2027\, and 50% by 2047.” (AI researchers (2023 expert survey)\, said Jan
  5\, 2024)\n\nHow we judge it: Our interpretation: YES if by end-2027 unai
 ded AI systems can do every task better and more cheaply than human worker
 s (the survey's HLMI definition).\n\nScorecard: https://sipoly.si/claim/ai
 -survey-2023-all-tasks-10pct-2027/
URL:https://sipoly.si/claim/ai-survey-2023-all-tasks-10pct-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-310
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: Deployed "last 10 yards" delivery robots: not before 2028 (Rod
 ney Brooks)
DESCRIPTION:“Lab demo: NET 2025 Deployed systems: NET 2028” (Rodney Bro
 oks\, said Jan 1\, 2018)\n\nHow we judge it: Our interpretation: correct i
 f no commercially deployed robot does vehicle-to-front-door package delive
 ry before 1 Jan 2028.\n\nScorecard: https://sipoly.si/claim/brooks-last-10
 -yards-deployed-net2028/
URL:https://sipoly.si/claim/brooks-last-10-yards-deployed-net2028/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-128
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: Multi-task physical-assistance robot for the elderly: not befo
 re 2028 (Rodney Brooks)
DESCRIPTION:“A robot that can provide physical assistance to the elderly 
 over multiple tasks (e.g.\, getting into and out of bed\, washing\, using 
 the toilet\, etc.) rather than just a point solution. NET 2028” (Rodney 
 Brooks\, said Jan 1\, 2018)\n\nHow we judge it: Correct if no such multi-t
 ask elder-assistance robot is available before 1 Jan 2028.\n\nScorecard: h
 ttps://sipoly.si/claim/brooks-elder-care-robot/
URL:https://sipoly.si/claim/brooks-elder-care-robot/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-120
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: Generally agreed "next big thing" beyond deep learning: by 202
 7 (Rodney Brooks)
DESCRIPTION:“Emergence of the generally agreed upon "next big thing" in A
 I beyond deep learning. NET 2023” (Rodney Brooks\, said Jan 1\, 2018)\n\
 nHow we judge it: Our interpretation: by end of 2027\, a widely agreed suc
 cessor paradigm to deep learning has emerged.\n\nScorecard: https://sipoly
 .si/claim/brooks-next-big-thing/
URL:https://sipoly.si/claim/brooks-next-big-thing/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-119
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: Driverless taxi service in 50 of the 100 biggest US cities: no
 t before 2028 (Rodney Brooks)
DESCRIPTION:“Such "taxi" service as above in 50 of the 100 biggest US cit
 ies. NET 2028” (Rodney Brooks\, said Jan 1\, 2018)\n\nHow we judge it: C
 orrect if no such service exists in 50 of the 100 largest US cities before
  1 Jan 2028.\n\nScorecard: https://sipoly.si/claim/brooks-taxi-50-cities/
URL:https://sipoly.si/claim/brooks-taxi-50-cities/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-436
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20271231
DTEND;VALUE=DATE:20280101
SUMMARY:Due: AI will outperform humans at driving a truck by 2027 (median) 
 (ML researchers (2016 expert survey))
DESCRIPTION:“Researchers predict AI will outperform humans in many activi
 ties in the next ten years\, such as translating languages (by 2024)\, wri
 ting high-school essays (by 2026)\, driving a truck (by 2027)\, working in
  retail (by 2031)\, writing a bestselling book (by 2049)\, and working as 
 a surgeon (by 2053).” (ML researchers (2016 expert survey)\, said May 24
 \, 2017)\n\nHow we judge it: Our interpretation: by end-2027\, driverless 
 trucks operate commercially at scale on public roads with safety at or bet
 ter than human drivers.\n\nScorecard: https://sipoly.si/claim/ml-survey-20
 16-truck-2027/
URL:https://sipoly.si/claim/ml-survey-2016-truck-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-411
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20280101
DTEND;VALUE=DATE:20280102
SUMMARY:Due: ~25%: 80%-reliability one-month task horizons by the start of 
 2028 (Ryan Greenblatt)
DESCRIPTION:“We're pretty unlikely (25%?) to see 80% reliability at 1 mon
 th tasks by the start of 2028.” (Ryan Greenblatt\, said Aug 2025)\n\nHow
  we judge it: Our interpretation: YES if by 1 Jan 2028 METR (or an equival
 ent evaluation) reports an 80%-reliability time horizon of at least one wo
 rk-month (~167 hours).\n\nScorecard: https://sipoly.si/claim/greenblatt-1-
 month-80pct-2028-25pct/
URL:https://sipoly.si/claim/greenblatt-1-month-80pct-2028-25pct/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-317
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20280110
DTEND;VALUE=DATE:20280111
SUMMARY:Due: "A 50% chance of AGI in the next 3 years" (from Jan 2025) (Sha
 ne Legg)
DESCRIPTION:“Of course this now means a 50% chance of AGI in the next 3 y
 ears!” (Shane Legg\, said Jan 10\, 2025)\n\nHow we judge it: Our interpr
 etation: by 10 Jan 2028\, a system exists that Google DeepMind or broad ex
 pert consensus accepts as AGI (able to do the cognitive tasks people typic
 ally can).\n\nScorecard: https://sipoly.si/claim/legg-agi-50pct-by-2028/
URL:https://sipoly.si/claim/legg-agi-50pct-by-2028/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-735
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20280131
DTEND;VALUE=DATE:20280201
SUMMARY:Due: US real GDP growth in 2027 above 3.3%\, ">50% higher than 2026
 " (Elon Musk)
DESCRIPTION:“My guess for real GDP growth next year is >50% higher than 2
 026\, so >3.3%” (Elon Musk\, said Sep 30\, 2026)\n\nHow we judge it: Our
  interpretation: correct if the US Bureau of Economic Analysis's first est
 imate of real GDP growth for calendar 2027 (annual-average basis\, normall
 y published in late January 2028) is above 3.3%.\n\nScorecard: https://sip
 oly.si/claim/musk-us-gdp-2027-above-3-3pct/
URL:https://sipoly.si/claim/musk-us-gdp-2027-above-3-3pct/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-733
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20280330
DTEND;VALUE=DATE:20280331
SUMMARY:Due: "Some form of AGI in the next six to 18 months" (Sep 2026) (Da
 n Schulman)
DESCRIPTION:“I think we get to some form of AGI in the next six to 18 mon
 ths. I think it’s right on us.” (Dan Schulman\, said Sep 30\, 2026)\n\
 nHow we judge it: Our interpretation: correct if by 30 Mar 2028 (18 months
  after the statement) an AI system is broadly recognized by independent ex
 perts as AGI\, i.e. able to do most economically valuable cognitive work a
 t the level of a skilled professional across domains. A developer's own AG
 I claim does not count on its own.\n\nScorecard: https://sipoly.si/claim/s
 chulman-agi-6-18-months/
URL:https://sipoly.si/claim/schulman-agi-6-18-months/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-123
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20280501
DTEND;VALUE=DATE:20280502
SUMMARY:Due: 30% of IBM back-office roles replaced over five years (Arvind 
 Krishna)
DESCRIPTION:“I could easily see 30% of that getting replaced by AI and au
 tomation over a five-year period.” (Arvind Krishna\, said May 1\, 2023)\
 n\nHow we judge it: By May 2028\, ~30% (~7\,800) of those IBM roles replac
 ed by AI/automation\, per IBM disclosures or credible reporting.\n\nScorec
 ard: https://sipoly.si/claim/krishna-30pct-back-office/
URL:https://sipoly.si/claim/krishna-30pct-back-office/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-410
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20280630
DTEND;VALUE=DATE:20280701
SUMMARY:Due: METR 50% time horizons of about two weeks around the start of 
 2028 (Ryan Greenblatt)
DESCRIPTION:“I expect doubling times of around 170 days on METR's task su
 ite (or similar tasks) [3] over the next 2 years or so which implies we'll
  be hitting 2 week 50% reliability horizon lengths around the start of 202
 8.” (Ryan Greenblatt\, said Aug 2025)\n\nHow we judge it: Our interpreta
 tion: correct if METR's best published 50% time horizon first reaches abou
 t 2 work-weeks (~80 hours) between Jul 2027 and Jun 2028\; wrong if it hap
 pens before Jul 2027 or is not reached by Jun 2028.\n\nScorecard: https://
 sipoly.si/claim/greenblatt-2-week-horizon-2028/
URL:https://sipoly.si/claim/greenblatt-2-week-horizon-2028/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-694
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20280630
DTEND;VALUE=DATE:20280701
SUMMARY:Due: Generative AI will reduce Amazon's total corporate workforce "
 in the next few years" (Amazon)
DESCRIPTION:“It’s hard to know exactly where this nets out over time\, 
 but in the next few years\, we expect that this will reduce our total corp
 orate workforce as we get efficiency gains from using AI extensively acros
 s the company.” (Amazon\, said Jun 2025)\n\nHow we judge it: Our interpr
 etation: Amazon's corporate (non-warehouse) headcount falls within about t
 hree years (by mid-2028)\, with AI cited as a driver.\n\nScorecard: https:
 //sipoly.si/claim/amazon-ai-shrinks-corporate-workforce/
URL:https://sipoly.si/claim/amazon-ai-shrinks-corporate-workforce/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-422
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20280630
DTEND;VALUE=DATE:20280701
SUMMARY:Due: Bet: AI will do at least 8 of Marcus's 10 hard tasks by the en
 d of 2027 (Miles Brundage)
DESCRIPTION:“If there exist AI systems that can perform 8 of the 10 tasks
  below by the end of 2027\, as determined by our panel of judges\, Gary wi
 ll donate $2\,000 to a charity of Miles’ choice” (Miles Brundage\, sai
 d Dec 30\, 2024)\n\nHow we judge it: Our interpretation: correct for Brund
 age if the agreed judging panel finds AI can do at least 8 of the 10 tasks
  by end-2027.\n\nScorecard: https://sipoly.si/claim/brundage-2027-bet-8-of
 -10-tasks/
URL:https://sipoly.si/claim/brundage-2027-bet-8-of-10-tasks/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-421
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20280630
DTEND;VALUE=DATE:20280701
SUMMARY:Due: Bet (10:1 odds): AI will not do 8 of 10 hard tasks by the end 
 of 2027 (Gary Marcus)
DESCRIPTION:“If there exist AI systems that can perform 8 of the 10 tasks
  below by the end of 2027\, as determined by our panel of judges\, Gary wi
 ll donate $2\,000 to a charity of Miles’ choice\; if AI can do fewer tha
 n 8\, Miles will donate $20\,000 to a charity of Gary’s choice.” (Gary
  Marcus\, said Dec 30\, 2024)\n\nHow we judge it: Our interpretation: corr
 ect for Marcus if the agreed judging panel finds AI can do fewer than 8 of
  the 10 tasks by end-2027\; resolved by the panel's ruling.\n\nScorecard: 
 https://sipoly.si/claim/marcus-2027-bet-fewer-than-8-tasks/
URL:https://sipoly.si/claim/marcus-2027-bet-fewer-than-8-tasks/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-121
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20280731
DTEND;VALUE=DATE:20280801
SUMMARY:Due: "There will be no programmers in five years" (Emad Mostaque)
DESCRIPTION:“There will be no programmers in five years.” (Emad Mostaqu
 e\, said Jul 3\, 2023)\n\nHow we judge it: Our interpretation: by Jul 2028
 \, professional software developer employment has largely disappeared (e.g
 . down 90%+).\n\nScorecard: https://sipoly.si/claim/mostaque-no-programmer
 s/
URL:https://sipoly.si/claim/mostaque-no-programmers/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-729
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20280923
DTEND;VALUE=DATE:20280924
SUMMARY:Due: A "country of geniuses in the data center" within "one or two 
 years\, maybe less" (Sep 2026) (Dario Amodei)
DESCRIPTION:“The trajectory only needs to continue for a tiny period long
 er\, one or two years\, maybe less\, to reach what I've called a country o
 f geniuses in the data center.” (Dario Amodei\, said Sep 23\, 2026)\n\nH
 ow we judge it: Our interpretation: correct if by 23 Sep 2028 independent 
 evaluators (not only Anthropic) judge that AI systems meet Amodei's own de
 scription: models smarter than top human experts across most fields\, able
  to work autonomously on tasks lasting days or weeks\, and run as many par
 allel copies.\n\nScorecard: https://sipoly.si/claim/amodei-country-of-geni
 uses-1-2y/
URL:https://sipoly.si/claim/amodei-country-of-geniuses-1-2y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-722
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20281029
DTEND;VALUE=DATE:20281030
SUMMARY:Due: AGI "likely to emerge in the next four years" (from Oct 2024) 
 (Thomas L. Friedman)
DESCRIPTION:“the birth of artificial general intelligence\, or A.G.I.\, w
 hich is likely to emerge in the next four years” (Thomas L. Friedman\, s
 aid Oct 29\, 2024)\n\nHow we judge it: Our interpretation: correct if by 2
 9 Oct 2028 an AI system is broadly recognized by independent experts as AG
 I (matching or beating humans across most cognitive work)\; a developer's 
 own claim does not count on its own.\n\nScorecard: https://sipoly.si/claim
 /friedman-agi-next-four-years/
URL:https://sipoly.si/claim/friedman-agi-next-four-years/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-354
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20281109
DTEND;VALUE=DATE:20281110
SUMMARY:Due: Within five years you will "simply tell your device" what you 
 want instead of using apps (Bill Gates)
DESCRIPTION:“In the next five years\, this will change completely. You wo
 n’t have to use different apps for different tasks. You’ll simply tell
  your device\, in everyday language\, what you want to do.” (Bill Gates\
 , said Nov 9\, 2023)\n\nHow we judge it: Our interpretation: by 9 Nov 2028
 \, mainstream consumers mostly accomplish multi-app tasks by telling an AI
  agent in natural language\, rather than opening separate apps.\n\nScoreca
 rd: https://sipoly.si/claim/gates-agents-five-years-2023/
URL:https://sipoly.si/claim/gates-agents-five-years-2023/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-728
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20281231
DTEND;VALUE=DATE:20290101
SUMMARY:Due: About a 50% chance that AI companies automate the whole AI res
 earch process by end of 2028 (Daniel Kokotajlo)
DESCRIPTION:“They are already starting to work on training AIs to do the 
 whole research process\, not just the coding. It’s unclear when they wil
 l succeed but my team and I think it could happen any year now. I personal
 ly would guess about 50% chance by end of 2028.” (Daniel Kokotajlo\, sai
 d Sep 30\, 2026)\n\nHow we judge it: Our interpretation: YES if by 31 Dec 
 2028 a frontier AI company has automated essentially the whole AI research
  and development process (AI systems\, not human researchers\, generate\, 
 run and interpret most of the research behind new frontier models\, with h
 umans mainly overseeing)\, as credibly reported by the company or independ
 ent evaluators. Stated probability 50%\, so it is scored on calibration.\n
 \nScorecard: https://sipoly.si/claim/kokotajlo-ai-rnd-automation-50pct-202
 8/
URL:https://sipoly.si/claim/kokotajlo-ai-rnd-automation-50pct-2028/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-711
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20281231
DTEND;VALUE=DATE:20290101
SUMMARY:Due: 1.6% of recently approved drug sales in 2027 will come from AI
 -discovered drugs (5% by 2030) (LEAP expert panel)
DESCRIPTION:“Experts predict that 1.6% of recently approved drug sales in
  2027 will come from AI-discovered drugs” (LEAP expert panel\, said Nov 
 10\, 2025)\n\nHow we judge it: Our interpretation: correct if credible 202
 7 analyses put the share of sales of recently approved drugs attributable 
 to AI-discovered drugs at 0.8-3%.\n\nScorecard: https://sipoly.si/claim/le
 ap-ai-drug-sales-1-6pct-2027/
URL:https://sipoly.si/claim/leap-ai-drug-sales-1-6pct-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-414
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20281231
DTEND;VALUE=DATE:20290101
SUMMARY:Due: AI will use 4% of US electricity in 2027 (7% in 2030) (LEAP ex
 pert panel)
DESCRIPTION:“The median expert predicts that 4% of U.S. electricity consu
 mption will be used for training and deploying AI systems in 2027.” (LEA
 P expert panel\, said Nov 10\, 2025)\n\nHow we judge it: Our interpretatio
 n: correct if credible 2027 estimates (EIA\, LBNL\, IEA\, Epoch) put AI tr
 aining plus inference at 3-5% of US electricity consumption.\n\nScorecard:
  https://sipoly.si/claim/leap-ai-electricity-4pct-2027/
URL:https://sipoly.si/claim/leap-ai-electricity-4pct-2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-348
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20281231
DTEND;VALUE=DATE:20290101
SUMMARY:Due: "Pretty confident" of AI that makes more significant discoveri
 es in 2028 and beyond (OpenAI)
DESCRIPTION:“In 2028 and beyond\, we are pretty confident we will have sy
 stems that can make more significant discoveries” (OpenAI\, said Nov 6\,
  2025)\n\nHow we judge it: Our interpretation: by the end of 2028\, an AI 
 system is credited with a significant scientific discovery (publishable in
  a top venue on its own merits\, with AI as the main contributor).\n\nScor
 ecard: https://sipoly.si/claim/openai-significant-discoveries-2028/
URL:https://sipoly.si/claim/openai-significant-discoveries-2028/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-349
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20281231
DTEND;VALUE=DATE:20290101
SUMMARY:Due: A fully automated "legitimate AI researcher" by 2028 (OpenAI)
DESCRIPTION:“internally\, OpenAI is tracking toward achieving an intern-l
 evel research assistant by September 2026 and a fully automated “legitim
 ate AI researcher” by 2028” (OpenAI\, said Oct 28\, 2025)\n\nHow we ju
 dge it: Our interpretation: by 31 Dec 2028\, OpenAI has a system that auto
 nomously carries out larger research projects\, publicly demonstrated or c
 redibly described.\n\nScorecard: https://sipoly.si/claim/openai-legit-ai-r
 esearcher-2028/
URL:https://sipoly.si/claim/openai-legit-ai-researcher-2028/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-705
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20281231
DTEND;VALUE=DATE:20290101
SUMMARY:Due: 33% of enterprise software apps will include agentic AI by 202
 8\; 15% of daily work decisions autonomous (Gartner)
DESCRIPTION:“Gartner predicts at least 15% of day-to-day work decisions w
 ill be made autonomously through agentic AI by 2028\, up from 0% in 2024. 
 In addition\, 33% of enterprise software applications will include agentic
  AI by 2028\, up from less than 1% in 2024.” (Gartner\, said Jun 25\, 20
 25)\n\nHow we judge it: Our interpretation: correct if Gartner's own (or c
 omparable analyst) 2028 data show at least 30% of enterprise software apps
  include agentic AI and at least 15% of day-to-day work decisions are made
  autonomously.\n\nScorecard: https://sipoly.si/claim/gartner-agentic-33pct
 -apps-2028/
URL:https://sipoly.si/claim/gartner-agentic-33pct-apps-2028/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-361
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20281231
DTEND;VALUE=DATE:20290101
SUMMARY:Due: 50/50 that AI does small-business taxes end-to-end by 2028 (Dw
 arkesh Patel)
DESCRIPTION:“AI can do taxes end-to-end for my small business as well as 
 a competent general manager could in a week: including chasing down all th
 e receipts on different websites\, finding all the missing pieces\, emaili
 ng back and forth with anyone we need to hassle for invoices\, filling out
  the form\, and sending it to the IRS: 2028” (Dwarkesh Patel\, said Jun 
 2\, 2025)\n\nHow we judge it: Our interpretation: YES if by 31 Dec 2028 a 
 generally available AI agent can do this end-to-end task as described.\n\n
 Scorecard: https://sipoly.si/claim/dwarkesh-ai-taxes-2028/
URL:https://sipoly.si/claim/dwarkesh-ai-taxes-2028/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-325
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20281231
DTEND;VALUE=DATE:20290101
SUMMARY:Due: 75% chance of AGI (replacing the majority of human jobs) by en
 d of 2028 (Andrew Critch)
DESCRIPTION:“p(AGI by eoy 2028) = 75%” (Andrew Critch\, said Feb 6\, 20
 25)\n\nHow we judge it: Our interpretation: AI that\, at runtime and for l
 ess than a human costs\, can replace humans in the power-weighted majority
  of jobs exists by 31 Dec 2028.\n\nScorecard: https://sipoly.si/claim/crit
 ch-agi-2028-75pct/
URL:https://sipoly.si/claim/critch-agi-2028-75pct/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-365
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20281231
DTEND;VALUE=DATE:20290101
SUMMARY:Due: Individual training clusters costing hundreds of billions of d
 ollars by 2028 (Leopold Aschenbrenner)
DESCRIPTION:“We’re on the path to individual training clusters costing 
 $100s of billions by 2028—clusters requiring power equivalent to a small
 /medium US state and more expensive than the International Space Station.
 ” (Leopold Aschenbrenner\, said Jun 4\, 2024)\n\nHow we judge it: Our in
 terpretation: by 31 Dec 2028\, at least one single training cluster (one c
 ampus) costs $100B or more\, per credible reporting or trackers.\n\nScorec
 ard: https://sipoly.si/claim/aschenbrenner-100b-clusters-2028/
URL:https://sipoly.si/claim/aschenbrenner-100b-clusters-2028/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-443
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20281231
DTEND;VALUE=DATE:20290101
SUMMARY:Due: At least 50%: AI builds a payment site from scratch\, writes a
  hit-indistinguishable song and fine-tunes an LLM by itself by 2028 (AI re
 searchers (2023 expert survey))
DESCRIPTION:“The aggregate forecasts give at least a 50% chance of AI sys
 tems achieving several milestones by 2028\, including autonomously constru
 cting a payment processing site from scratch\, creating a song indistingui
 shable from a new song by a popular musician\, and autonomously downloadin
 g and fine-tuning a large language model.” (AI researchers (2023 expert 
 survey)\, said Jan 5\, 2024)\n\nHow we judge it: Our interpretation: YES i
 f by end-2028 all three milestones are credibly demonstrated: an AI autono
 mously builds a working payment-processing site from scratch\, generates a
  song judged indistinguishable from a new release by a popular artist\, an
 d autonomously downloads and fine-tunes an LLM.\n\nScorecard: https://sipo
 ly.si/claim/ai-survey-2023-milestones-2028/
URL:https://sipoly.si/claim/ai-survey-2023-milestones-2028/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-408
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20290101
DTEND;VALUE=DATE:20290102
SUMMARY:Due: ~15%: AI capable of fully automating AI R&D by the start of 20
 29 (Ryan Greenblatt)
DESCRIPTION:“I now would put ~15% probability by the start of 2029 and a 
 45% chance by the start of 2033.” (Ryan Greenblatt\, said Aug 2025)\n\nH
 ow we judge it: Our interpretation: YES if by 1 Jan 2029 an AI system exis
 ts that could fully automate AI R&D (do all the work of a frontier lab's r
 esearch staff).\n\nScorecard: https://sipoly.si/claim/greenblatt-ai-rd-aut
 omation-2029-15pct/
URL:https://sipoly.si/claim/greenblatt-ai-rd-automation-2029-15pct/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-374
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20290120
DTEND;VALUE=DATE:20290121
SUMMARY:Due: AGI "will probably get developed during this president's term"
  (Sam Altman)
DESCRIPTION:“AGI will probably get developed during this president's term
 \, and getting that right seems really important.” (Sam Altman\, said Ja
 n 6\, 2025)\n\nHow we judge it: Our interpretation: AGI (as broadly recogn
 ized\, not merely self-declared) is developed before 20 Jan 2029.\n\nScore
 card: https://sipoly.si/claim/altman-agi-this-term/
URL:https://sipoly.si/claim/altman-agi-this-term/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-124
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20290121
DTEND;VALUE=DATE:20290122
SUMMARY:Due: Stargate: $500B of US AI infrastructure over four years (OpenA
 I)
DESCRIPTION:“The Stargate Project is a new company which intends to inves
 t $500 billion over the next four years building new AI infrastructure for
  OpenAI in the United States.” (OpenAI\, said Jan 21\, 2025)\n\nHow we j
 udge it: Our interpretation: credible disclosures show roughly $500B inves
 ted in Stargate US AI infrastructure by 21 Jan 2029.\n\nScorecard: https:/
 /sipoly.si/claim/stargate-500b-four-years/
URL:https://sipoly.si/claim/stargate-500b-four-years/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-350
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20290130
DTEND;VALUE=DATE:20290131
SUMMARY:Due: Bet: AI will not get an A on 5 of his 6 latest midterms by Jan
 uary 2029 (Bryan Caplan)
DESCRIPTION:“If the AI gets an A on at least 5 of out 6 of those exams us
 ing same grading scale as his students\, then Bryan owes Matthew $500.” 
 (Bryan Caplan\, said Jan 19\, 2023)\n\nHow we judge it: Our interpretation
 : Caplan is right if an AI does not get an A (A- counts) on at least 5 of 
 his 6 most recent midterms by 30 Jan 2029\, under the bet's terms.\n\nScor
 ecard: https://sipoly.si/claim/caplan-ai-no-a-midterms-2029/
URL:https://sipoly.si/claim/caplan-ai-no-a-midterms-2029/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-369
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20290319
DTEND;VALUE=DATE:20290320
SUMMARY:Due: AI will do well on any specified test set ("AGI") within 5 yea
 rs (Jensen Huang)
DESCRIPTION:“If we specified AGI to be something very specific\, a set of
  tests where a software program can do very well — or maybe 8% better th
 an most people — I believe we will get there within 5 years” (Jensen H
 uang\, said Mar 19\, 2024)\n\nHow we judge it: Our interpretation: by 19 M
 ar 2029\, AI systems score better than most humans on essentially any stan
 dardized human test (professional\, academic and logic exams).\n\nScorecar
 d: https://sipoly.si/claim/huang-agi-tests-5-years/
URL:https://sipoly.si/claim/huang-agi-tests-5-years/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-328
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20290529
DTEND;VALUE=DATE:20290530
SUMMARY:Due: AI will displace social drinking within five years (Jonathan R
 oss)
DESCRIPTION:“Prediction: AI will displace social drinking within 5 years
 ” (Jonathan Ross\, said May 29\, 2024)\n\nHow we judge it: Our interpret
 ation: by 29 May 2029\, credible data show AI social assistants (e.g. earb
 uds) have measurably reduced social alcohol consumption\, e.g. surveys att
 ributing reduced drinking to AI use.\n\nScorecard: https://sipoly.si/claim
 /ross-ai-displaces-social-drinking/
URL:https://sipoly.si/claim/ross-ai-displaces-social-drinking/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-366
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20290611
DTEND;VALUE=DATE:20290612
SUMMARY:Due: In five years there will be more software engineers than today
 \, not fewer (François Chollet)
DESCRIPTION:“Sure. In five years\, there will be more software engineers 
 than there are today\, not fewer.” (François Chollet\, said Jun 11\, 20
 24)\n\nHow we judge it: Our interpretation: correct if US employment of so
 ftware developers/engineers (BLS OES/CPS) in mid-2029 is above its 2024 le
 vel.\n\nScorecard: https://sipoly.si/claim/chollet-more-swes-in-5y/
URL:https://sipoly.si/claim/chollet-more-swes-in-5y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-367
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20290621
DTEND;VALUE=DATE:20290622
SUMMARY:Due: AGI "one to 10 times smarter than humans" in the next three to
  five years (Masayoshi Son)
DESCRIPTION:“Son said this tech is likely to be one to 10 times smarter t
 han humans and will arrive in the next three-to-five years\, earlier than 
 he had anticipated.” (Masayoshi Son\, said Jun 21\, 2024)\n\nHow we judg
 e it: Our interpretation: AGI (AI at least as capable as humans across mos
 t cognitive tasks) exists by 21 Jun 2029.\n\nScorecard: https://sipoly.si/
 claim/son-agi-3-5-years/
URL:https://sipoly.si/claim/son-agi-3-5-years/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-383
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20290630
DTEND;VALUE=DATE:20290701
SUMMARY:Due: Data-center AI accelerator market to reach $500 billion by 202
 8 (AMD)
DESCRIPTION:“Looking ahead\, we see the data center AI accelerator market
  growing to $500 billion by 2028.” (AMD\, said Oct 10\, 2024)\n\nHow we 
 judge it: Our interpretation: credible market estimates (e.g. from AMD\, N
 vidia filings\, analysts) put 2028 data-center AI accelerator revenue at $
 500B or more.\n\nScorecard: https://sipoly.si/claim/amd-ai-accelerator-tam
 -500b-2028/
URL:https://sipoly.si/claim/amd-ai-accelerator-tam-500b-2028/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-125
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20290704
DTEND;VALUE=DATE:20290705
SUMMARY:Due: Binding regulation on the most powerful AI model developers (U
 K Labour Party / UK Government)
DESCRIPTION:“Labour will ensure the safe development and use of AI models
  by introducing binding regulation on the handful of companies developing 
 the most powerful AI models” (UK Labour Party / UK Government\, said Jun
  13\, 2024)\n\nHow we judge it: Our interpretation: binding legislation ap
 plying to frontier AI developers enacted in the UK during this Parliament 
 (we use 4 Jul 2029).\n\nScorecard: https://sipoly.si/claim/uk-labour-bindi
 ng-ai-regulation/
URL:https://sipoly.si/claim/uk-labour-binding-ai-regulation/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-724
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20291031
DTEND;VALUE=DATE:20291101
SUMMARY:Due: AI systems improving their own code in ~5 years\; 80-90% of ex
 pert ability in every field in 6-8 years (Eric Schmidt)
DESCRIPTION:“In the industry it is believed that somewhere around five ye
 ars\, no one knows exactly\, the systems will begin to be able to write th
 eir own code\, that is\, they literally will take their code and make it b
 etter.” (Eric Schmidt\, said Oct 2024)\n\nHow we judge it: Our interpret
 ation: correct if by 31 Oct 2029 frontier AI systems are documented (by a 
 lab or independent evaluators) as substantially improving their own code o
 r training process\, i.e. recursive self-improvement in practice.\n\nScore
 card: https://sipoly.si/claim/schmidt-recursive-code-5y-experts-6-8y/
URL:https://sipoly.si/claim/schmidt-recursive-code-5y-experts-6-8y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-446
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20291231
DTEND;VALUE=DATE:20300101
SUMMARY:Due: Median 2029 for full coding automation (as of Feb 2026) (Danie
 l Kokotajlo)
DESCRIPTION:“Mid-2028 is earlier than Daniel’s current median predictio
 n for full coding automation (2029)\, but the 2-year takeoff to superintel
 ligence is slower than his median takeoff speed of ~1 year.” (Daniel Kok
 otajlo\, said Feb 12\, 2026)\n\nHow we judge it: Our interpretation: YES i
 f by end-2029 AI can fully automate coding (do any coding task the best AG
 I-company engineer does\, faster and cheaper).\n\nScorecard: https://sipol
 y.si/claim/kokotajlo-full-coding-automation-2029-median/
URL:https://sipoly.si/claim/kokotajlo-full-coding-automation-2029-median/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-319
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20291231
DTEND;VALUE=DATE:20300101
SUMMARY:Due: No halving of US labor-force participation and no 15% GDP grow
 th year by 2029 (Matthew Barnett)
DESCRIPTION:“I don't expect the US labor force participation rate to fall
  by more than 50% from its current level by 2029\, or for US GDP growth to
  surpass 15% for any year in 2025-2029.” (Matthew Barnett\, said May 18\
 , 2025)\n\nHow we judge it: Our interpretation: correct if through 2029 th
 e US labor force participation rate never falls below half its May 2025 le
 vel (~62.4%) and no calendar year 2025-2029 has real US GDP growth above 1
 5%.\n\nScorecard: https://sipoly.si/claim/barnett-no-lfpr-collapse-2029/
URL:https://sipoly.si/claim/barnett-no-lfpr-collapse-2029/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-418
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20291231
DTEND;VALUE=DATE:20300101
SUMMARY:Due: LLM hallucinations "much less of a problem\, certainly by 2029
 " (Ray Kurzweil)
DESCRIPTION:“LLM hallucinations [where they create nonsensical or inaccur
 ate outputs] will become much less of a problem\, certainly by 2029” (Ra
 y Kurzweil\, said Jun 29\, 2024)\n\nHow we judge it: Our interpretation: b
 y end-2029\, frontier models' hallucination rates on standard factuality b
 enchmarks (e.g. SimpleQA-style or OpenAI/Vectara hallucination leaderboard
 s) are a small fraction (≤ one-fifth) of 2024 levels\, and hallucination
  is no longer named a top barrier to deployment.\n\nScorecard: https://sip
 oly.si/claim/kurzweil-hallucinations-solved-2029/
URL:https://sipoly.si/claim/kurzweil-hallucinations-solved-2029/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-122
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20291231
DTEND;VALUE=DATE:20300101
SUMMARY:Due: Human-level intelligence and AGI by 2029 (Ray Kurzweil)
DESCRIPTION:“So 2029\, both for human-level intelligence and for artifici
 al general intelligence (AGI) – which is a little bit different.” (Ray
  Kurzweil\, said Jun 29\, 2024)\n\nHow we judge it: Our interpretation: by
  31 Dec 2029\, AI matches the most skilled humans in most domains.\n\nScor
 ecard: https://sipoly.si/claim/kurzweil-2029/
URL:https://sipoly.si/claim/kurzweil-2029/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-671
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20291231
DTEND;VALUE=DATE:20300101
SUMMARY:Due: By 2029\, AI "probably smarter than all humans combined" (Elon
  Musk)
DESCRIPTION:“By 2029\, AI is probably smarter than all humans combined.
 ” (Elon Musk\, said Mar 13\, 2024)\n\nHow we judge it: Our interpretatio
 n: by 31 Dec 2029\, a single AI system (or coordinated AI collective) out-
 thinks humanity's combined intellectual output\, e.g. it can do essentiall
 y all cognitive work faster and better than all human experts together.\n\
 nScorecard: https://sipoly.si/claim/musk-2029-smarter-than-all-humans/
URL:https://sipoly.si/claim/musk-2029-smarter-than-all-humans/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-665
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20291231
DTEND;VALUE=DATE:20300101
SUMMARY:Due: AI with an ongoing existence at the level of a mouse: not befo
 re 2030 (Rodney Brooks)
DESCRIPTION:“An AI system with an ongoing existence (no day is the repeat
  of another day as it currently is for all AI systems) at the level of a m
 ouse. NET 2030” (Rodney Brooks\, said Jan 1\, 2018)\n\nHow we judge it: 
 Our interpretation: correct if before 1 Jan 2030 no AI system is generally
  accepted as having a continuous\, learning\, mouse-level existence.\n\nSc
 orecard: https://sipoly.si/claim/brooks-mouse-level-ai-net2030/
URL:https://sipoly.si/claim/brooks-mouse-level-ai-net2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-309
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20291231
DTEND;VALUE=DATE:20300101
SUMMARY:Due: Dexterous robot hands generally available: not before 2030 (ho
 pefully by 2040) (Rodney Brooks)
DESCRIPTION:“Dexterous robot hands generally available. NET 2030 BY 2040 
 (I hope!)” (Rodney Brooks\, said Jan 1\, 2018)\n\nHow we judge it: Our i
 nterpretation: correct if articulated\, human-like dexterous robot hands a
 re not generally available as deployed products before 1 Jan 2030.\n\nScor
 ecard: https://sipoly.si/claim/brooks-dexterous-hands-net2030/
URL:https://sipoly.si/claim/brooks-dexterous-hands-net2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-307
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20291231
DTEND;VALUE=DATE:20300101
SUMMARY:Due: Home-navigating robot as an expensive product: not before 2030
  (Rodney Brooks)
DESCRIPTION:“A robot that can navigate around just about any US home\, wi
 th its steps\, its clutter\, its narrow pathways between furniture\, etc. 
 Lab demo: NET 2026 Expensive product: NET 2030” (Rodney Brooks\, said Ja
 n 1\, 2018)\n\nHow we judge it: Our interpretation: correct if no robot ab
 le to navigate almost any US home (steps\, clutter\, narrow paths) is sold
  as a product\, at any price\, before 1 Jan 2030.\n\nScorecard: https://si
 poly.si/claim/brooks-home-robot-expensive-net2030/
URL:https://sipoly.si/claim/brooks-home-robot-expensive-net2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-346
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20291231
DTEND;VALUE=DATE:20300101
SUMMARY:Due: Long Bet: no computer will pass the Turing Test by 2029 (Mitch
  Kapor)
DESCRIPTION:“By 2029 no computer - or "machine intelligence" - will have 
 passed the Turing Test.” (Mitch Kapor\, said 2002)\n\nHow we judge it: O
 ur interpretation: as defined in the Long Bets terms: correct if no machin
 e passes the Kapor-Kurzweil Turing Test protocol by the end of 2029.\n\nSc
 orecard: https://sipoly.si/claim/kapor-no-turing-test-2029/
URL:https://sipoly.si/claim/kapor-no-turing-test-2029/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-84
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20300123
DTEND;VALUE=DATE:20300124
SUMMARY:Due: Current LLM paradigm has a shelf life of "probably three to fi
 ve years" (Yann LeCun)
DESCRIPTION:“I think the shelf life of the current [LLM] paradigm is fair
 ly short\, probably three to five years” (Yann LeCun\, said Jan 23\, 202
 5)\n\nHow we judge it: Our interpretation: by 23 Jan 2030\, LLMs are no lo
 nger the central component of leading AI systems.\n\nScorecard: https://si
 poly.si/claim/lecun-llm-shelf-life/
URL:https://sipoly.si/claim/lecun-llm-shelf-life/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-320
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20300311
DTEND;VALUE=DATE:20300312
SUMMARY:Due: Bet: AI produces Annals-quality number theory papers for under
  $100k each within 5 years (Tamay Besiroglu)
DESCRIPTION:“I bet @littmath that in 5 years AI will be able to produce A
 nnals-quality Number Theory papers at an inference budget at or below $100
 k/paper\, at 3:1 odds in my favor.” (Tamay Besiroglu\, said Mar 11\, 202
 5)\n\nHow we judge it: Our interpretation: by 11 Mar 2030\, AI produces nu
 mber-theory papers judged of Annals of Mathematics quality (per the bet te
 rms with Daniel Litt) at or below $100k inference cost per paper.\n\nScore
 card: https://sipoly.si/claim/besiroglu-annals-number-theory-2030/
URL:https://sipoly.si/claim/besiroglu-annals-number-theory-2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-388
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20300331
DTEND;VALUE=DATE:20300401
SUMMARY:Due: Long Bet: AI will not lift US productivity growth above 1.8%/y
 r in the 2020s (Robert J. Gordon)
DESCRIPTION:“Will robots and AI bring a new revival comparable to the pos
 t-1995 digital revival? A reason for doubt is that a doubling of the U.S. 
 stock of robots in the past decade failed to revive manufacturing producti
 vity growth” (Robert J. Gordon\, said 2021)\n\nHow we judge it: Our inte
 rpretation: Gordon is right if BLS private nonfarm business productivity g
 rows at an average of 1.8%/yr or less from 2020Q1 to 2029Q4.\n\nScorecard:
  https://sipoly.si/claim/gordon-no-productivity-revival-2029/
URL:https://sipoly.si/claim/gordon-no-productivity-revival-2029/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-387
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20300331
DTEND;VALUE=DATE:20300401
SUMMARY:Due: Long Bet: US productivity growth will average over 1.8%/yr in 
 2020-2029 (Erik Brynjolfsson)
DESCRIPTION:“Private Nonfarm business productivity growth will average ov
 er 1.8 percent per year from the first quarter (Q1) of 2020 to the last qu
 arter of 2029 (Q4).” (Erik Brynjolfsson\, said 2021)\n\nHow we judge it:
  Our interpretation: as in the bet terms\, BLS private nonfarm business la
 bor productivity grows at an average above 1.8%/yr from 2020Q1 to 2029Q4.\
 n\nScorecard: https://sipoly.si/claim/brynjolfsson-productivity-1-8pct-202
 9/
URL:https://sipoly.si/claim/brynjolfsson-productivity-1-8pct-2029/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-720
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20300522
DTEND;VALUE=DATE:20300523
SUMMARY:Due: "Almost guaranteed": a drop-in white-collar worker within five
  years\, "very likely in two" (May 2025) (Sholto Douglas)
DESCRIPTION:“But the one that I feel we're almost guaranteed to get—thi
 s is a strong statement to make—is one where at the very least\, you get
  a drop-in white collar worker at some point in the next five years. I thi
 nk it's very likely in two\, but it seems almost overdetermined in five.
 ” (Sholto Douglas\, said May 22\, 2025)\n\nHow we judge it: Our interpre
 tation: correct if by 22 May 2030 AI agents can be deployed as drop-in rep
 lacements for typical white-collar workers (doing essentially any white-co
 llar job end to end)\, as broadly recognized by independent observers\; a 
 developer's own claim does not count on its own.\n\nScorecard: https://sip
 oly.si/claim/sholto-white-collar-automatable-2028/
URL:https://sipoly.si/claim/sholto-white-collar-automatable-2028/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-673
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20300523
DTEND;VALUE=DATE:20300524
SUMMARY:Due: AI able to replace more than half of humans "within 5 years" (
 from May 2025) (Scott Alexander)
DESCRIPTION:“I think AI will be able to replace >50% of humans within 5 y
 ears.” (Scott Alexander\, said May 23\, 2025)\n\nHow we judge it: Our in
 terpretation: by 23 May 2030\, AI can do the jobs of more than half of the
  human workforce at lower cost (capability\, not necessarily deployment).\
 n\nScorecard: https://sipoly.si/claim/scott-alexander-replace-half-5y/
URL:https://sipoly.si/claim/scott-alexander-replace-half-5y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-343
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20300528
DTEND;VALUE=DATE:20300529
SUMMARY:Due: AI could wipe out half of entry-level white-collar jobs and pu
 sh unemployment to 10-20% within 1-5 years (Dario Amodei)
DESCRIPTION:“AI could wipe out half of all entry-level white-collar jobs 
 — and spike unemployment to 10-20% in the next one to five years\, Amode
 i told us in an interview from his San Francisco office.” (Dario Amodei\
 , said May 28\, 2025)\n\nHow we judge it: Our interpretation: by 28 May 20
 30\, US unemployment (BLS U-3) reaches 10% or more at some point\, with AI
  widely credited\, or entry-level white-collar employment falls by about h
 alf. Hedged ("could").\n\nScorecard: https://sipoly.si/claim/amodei-white-
 collar-half-1-5y/
URL:https://sipoly.si/claim/amodei-white-collar-half-1-5y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-708
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20300701
DTEND;VALUE=DATE:20300702
SUMMARY:Due: Within five years AI will be able to do 80% of 80% of all jobs
  (Vinod Khosla)
DESCRIPTION:““Within the next five years\, any economically valuable jo
 b humans can do\, AI will be able to do 80% of it…80% of all jobs can be
  done by an AI.”” (Vinod Khosla\, said Jun 2025)\n\nHow we judge it: O
 ur interpretation: by mid-2030\, AI systems can perform about 80% of the t
 asks in about 80% of economically valuable jobs (capability\, not necessar
 ily deployment).\n\nScorecard: https://sipoly.si/claim/khosla-ai-80pct-of-
 jobs-5y/
URL:https://sipoly.si/claim/khosla-ai-80pct-of-jobs-5y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-371
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20300904
DTEND;VALUE=DATE:20300905
SUMMARY:Due: "In five years" unemployment levels never seen before: "Not ta
 lking about 10% ... but 99%" (Roman Yampolskiy)
DESCRIPTION:“In five years\, we’re looking at levels of unemployment we
 ’ve never seen before” (Roman Yampolskiy\, said Sep 4\, 2025)\n\nHow w
 e judge it: Our interpretation: correct if by Sep 2030 the US unemployment
  rate exceeds its post-1948 record (14.8%). Full credit for his headline "
 99%" would need near-total unemployment.\n\nScorecard: https://sipoly.si/c
 laim/yampolskiy-99pct-unemployment-5y/
URL:https://sipoly.si/claim/yampolskiy-99pct-unemployment-5y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-321
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20301231
DTEND;VALUE=DATE:20310101
SUMMARY:Due: Median 2030 for full coding automation (AI Futures Model) (Eli
  Lifland)
DESCRIPTION:“With my parameters\, it predicts a median of 2030 for full c
 oding automation” (Eli Lifland\, said Dec 31\, 2025)\n\nHow we judge it:
  Our interpretation: by 31 Dec 2030\, AI can fully automate the coding wor
 k of top AI-company engineers (the AI 2027 "superhuman coder" milestone).\
 n\nScorecard: https://sipoly.si/claim/lifland-coding-automation-2030/
URL:https://sipoly.si/claim/lifland-coding-automation-2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-413
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20301231
DTEND;VALUE=DATE:20310101
SUMMARY:Due: 20%: AI solves or substantially helps solve a Millennium Prize
  Problem by 2030 (LEAP expert panel)
DESCRIPTION:“Experts estimate a 10% chance that AI will solve or substant
 ially assist in solving a Millennium Prize Problem by 2027\, 2 up to 20% b
 y 2030” (LEAP expert panel\, said Nov 10\, 2025)\n\nHow we judge it: Our
  interpretation: YES if by 31 Dec 2030 a Millennium Prize Problem is accep
 ted as solved\, with AI credited as solving it or substantially assisting.
 \n\nScorecard: https://sipoly.si/claim/leap-millennium-prize-2030-20pct/
URL:https://sipoly.si/claim/leap-millennium-prize-2030-20pct/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-433
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20301231
DTEND;VALUE=DATE:20310101
SUMMARY:Due: By 2030 AI will autonomously fix issues\, implement features a
 nd solve hard scientific programming problems (Epoch AI)
DESCRIPTION:“By 2030\, on current trends\, AI will be able to autonomousl
 y fix issues\, implement features\, and solve difficult (but well-defined)
  scientific programming problems.” (Epoch AI\, said Sep 2025)\n\nHow we 
 judge it: Our interpretation: by end-2030\, AI agents routinely resolve re
 al-world software issues and implement features end to end without human c
 ode edits (e.g. SWE-bench-style benchmarks saturated\, plus documented pro
 duction use)\, and solve well-specified scientific programming tasks.\n\nS
 corecard: https://sipoly.si/claim/epoch-autonomous-swe-2030/
URL:https://sipoly.si/claim/epoch-autonomous-swe-2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-432
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20301231
DTEND;VALUE=DATE:20310101
SUMMARY:Due: Frontier training clusters will cost over $100B by 2030\, enab
 ling ~1e29 FLOP runs (Epoch AI)
DESCRIPTION:“On current trends\, the clusters used for training frontier 
 AI would cost over $100B by 2030. Such clusters could support training run
 s of about 10^29 FLOP” (Epoch AI\, said Sep 2025)\n\nHow we judge it: Ou
 r interpretation: correct if by end-2030 at least one frontier training cl
 uster has a reported cost above $100B\, and the largest known training run
  is within 10x of 1e29 FLOP.\n\nScorecard: https://sipoly.si/claim/epoch-1
 00b-clusters-1e29-2030/
URL:https://sipoly.si/claim/epoch-100b-clusters-1e29-2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-402
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20301231
DTEND;VALUE=DATE:20310101
SUMMARY:Due: Few if any drugs approved by 2030 will have benefited from tod
 ay's AI tools (Epoch AI)
DESCRIPTION:“we expect that few if any of the drugs approved for sale by 
 2030 will have benefited from today’s AI tools\, let alone those of 2030
 .” (Epoch AI\, said Sep 2025)\n\nHow we judge it: Our interpretation: co
 rrect if\, by end-2030\, no more than a handful (≤3) of FDA/EMA-approved
  drugs are credibly described as discovered or designed with 2025-era AI t
 ools.\n\nScorecard: https://sipoly.si/claim/epoch-few-ai-drugs-by-2030/
URL:https://sipoly.si/claim/epoch-few-ai-drugs-by-2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-401
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20301231
DTEND;VALUE=DATE:20310101
SUMMARY:Due: By 2030 many scientific fields will have AI assistants compara
 ble to today's coding assistants (Epoch AI)
DESCRIPTION:“By 2030\, we predict that many scientific domains will have 
 AI assistants comparable to coding assistants for software engineers today
 .” (Epoch AI\, said Sep 2025)\n\nHow we judge it: Our interpretation: by
  2030\, AI research assistants are in routine use (comparable to 2025 codi
 ng assistants) in most major scientific fields.\n\nScorecard: https://sipo
 ly.si/claim/epoch-science-assistants-2030/
URL:https://sipoly.si/claim/epoch-science-assistants-2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-431
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20301231
DTEND;VALUE=DATE:20310101
SUMMARY:Due: Hyperion data centre: 2 GW online by 2030\, scaling to 5 GW (M
 eta)
DESCRIPTION:“Gabriel says Meta plans to bring two gigawatts of data cente
 r capacity online by 2030 with Hyperion\, but that it would scale to five 
 gigawatts in several years.” (Meta\, said Jul 14\, 2025)\n\nHow we judge
  it: Our interpretation: Meta's Hyperion campus has at least 2 GW of data-
 centre capacity operational by 31 Dec 2030 (per Meta\, utility filings or 
 Epoch AI satellite tracking).\n\nScorecard: https://sipoly.si/claim/meta-h
 yperion-2gw-2030/
URL:https://sipoly.si/claim/meta-hyperion-2gw-2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-336
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20301231
DTEND;VALUE=DATE:20310101
SUMMARY:Due: UK to expand its AI Research Resource compute "by at least 20x
  by 2030" (UK Labour Party / UK Government)
DESCRIPTION:“Expand the capacity of the AI Research Resource ( AIRR ) by 
 at least 20x by 2030 - starting within 6 months.” (UK Labour Party / UK 
 Government\, said Jan 13\, 2025)\n\nHow we judge it: Our interpretation: b
 y 31 Dec 2030\, UK government reports AIRR capacity at least 20 times its 
 January 2025 level.\n\nScorecard: https://sipoly.si/claim/uk-airr-20x-by-2
 030/
URL:https://sipoly.si/claim/uk-airr-20x-by-2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-326
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20301231
DTEND;VALUE=DATE:20310101
SUMMARY:Due: Claims of AGI by 2030 are "laughable" (Christopher Manning)
DESCRIPTION:“I do not believe human-level AI (artificial superintelligenc
 e\, or the commonest sense of #AGI) is close at hand. AI has made breakthr
 oughs\, but the claim of AGI by 2030 is as laughable as claims of AGI by 1
 980 are in retrospect.” (Christopher Manning\, said Mar 14\, 2024)\n\nHo
 w we judge it: Our interpretation: correct if by 31 Dec 2030 no AI system 
 is broadly accepted as human-level AGI.\n\nScorecard: https://sipoly.si/cl
 aim/manning-agi-2030-laughable/
URL:https://sipoly.si/claim/manning-agi-2030-laughable/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-426
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20301231
DTEND;VALUE=DATE:20310101
SUMMARY:Due: Human-level AGI most likely in 2029-2030\, possibly as early a
 s 2027 (Ben Goertzel)
DESCRIPTION:“Goertzel suggested 2029 or 2030 could be the likeliest years
  when humanity will build the first AGI agent\, but that it could happen a
 s early as 2027.” (Ben Goertzel\, said Mar 2024)\n\nHow we judge it: Our
  interpretation: correct if a human-level AGI agent (broadly as capable as
  humans across cognitive tasks) is built by end-2030.\n\nScorecard: https:
 //sipoly.si/claim/goertzel-agi-2027-2030/
URL:https://sipoly.si/claim/goertzel-agi-2027-2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-699
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20301231
DTEND;VALUE=DATE:20310101
SUMMARY:Due: By 2030 AI will outcompete most professional mathematicians at
  research (Jacob Steinhardt)
DESCRIPTION:“In a domain like mathematics research where work can be chec
 ked automatically\, I’d predict that GPT 2030 will outcompete most profe
 ssional mathematicians.” (Jacob Steinhardt\, said Jun 7\, 2023)\n\nHow w
 e judge it: Our interpretation: by end-2030\, AI systems routinely produce
  publishable mathematical research at or above the level of a typical prof
 essional mathematician\, as judged by mainstream mathematicians.\n\nScorec
 ard: https://sipoly.si/claim/steinhardt-gpt2030-outcompete-mathematicians/
URL:https://sipoly.si/claim/steinhardt-gpt2030-outcompete-mathematicians/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-407
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20301231
DTEND;VALUE=DATE:20310101
SUMMARY:Due: "GPT 2030" will likely be superhuman at coding\, hacking and m
 ath (Jacob Steinhardt)
DESCRIPTION:“GPT 2030 will likely be superhuman at various specific tasks
 \, including coding\, hacking\, and math\, and potentially protein design
 ” (Jacob Steinhardt\, said Jun 7\, 2023)\n\nHow we judge it: Our interpr
 etation: by end-2030\, the best available AI system beats top human expert
 s on standard measures of coding (e.g. top competitive-programming ranks)\
 , offensive security\, and mathematics.\n\nScorecard: https://sipoly.si/cl
 aim/steinhardt-gpt2030-superhuman-coding-math/
URL:https://sipoly.si/claim/steinhardt-gpt2030-superhuman-coding-math/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-688
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20301231
DTEND;VALUE=DATE:20310101
SUMMARY:Due: ~15% probability of transformative AI by 2030 (Ajeya Cotra)
DESCRIPTION:“~15% probability by 2030 (a decrease of ~6 years from 2036).
 ” (Ajeya Cotra\, said Aug 2\, 2022)\n\nHow we judge it: Our interpretati
 on: YES if transformative AI (as defined in her report) exists by 31 Dec 2
 030.\n\nScorecard: https://sipoly.si/claim/cotra-tai-2030-15pct/
URL:https://sipoly.si/claim/cotra-tai-2030-15pct/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-674
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20301231
DTEND;VALUE=DATE:20310101
SUMMARY:Due: China to be "the world's primary AI innovation center" by 2030
  (China State Council)
DESCRIPTION:“Third\, by 2030\, China’s AI theories\, technologies\, and
  applications should achieve world-leading levels\, making China the world
 ’s primary AI innovation center” (China State Council\, said Jul 20\, 
 2017)\n\nHow we judge it: Our interpretation: by 31 Dec 2030\, China leads
  the world in AI research and frontier capability (e.g. top frontier model
 s and most highly-cited AI research)\, per major independent indices such 
 as the Stanford AI Index.\n\nScorecard: https://sipoly.si/claim/china-aidp
 -world-leader-2030/
URL:https://sipoly.si/claim/china-aidp-world-leader-2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-363
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20310918
DTEND;VALUE=DATE:20310919
SUMMARY:Due: "Real AI" (beyond today's chatbots) arrives in about 5 years (
 David Krueger)
DESCRIPTION:“No. We’re not there yet. But the real AI is coming. I thin
 k we’ve got about 5 years.” (David Krueger\, said Sep 18\, 2025)\n\nHo
 w we judge it: Our interpretation: AI that can broadly replace human cogni
 tive work arrives by about Sep 2030 (we allow a year either way).\n\nScore
 card: https://sipoly.si/claim/krueger-real-ai-5-years/
URL:https://sipoly.si/claim/krueger-real-ai-5-years/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-434
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20311231
DTEND;VALUE=DATE:20320101
SUMMARY:Due: ~30% of physics\, materials science and medicine papers will e
 ngage AI by 2030 (from 3% in 2022) (LEAP expert panel)
DESCRIPTION:“Experts predict a 10x increase (from 3% to ~30%) in AI-engag
 ed papers in Physics\, 10 Materials Science\, 11 and Medicine\, 12 between
  2022 and 2030.” (LEAP expert panel\, said Nov 10\, 2025)\n\nHow we judg
 e it: Our interpretation: correct if the Duede et al.-style measure of AI-
 engaged papers reaches roughly 20-30% in these fields in 2030.\n\nScorecar
 d: https://sipoly.si/claim/leap-ai-engaged-papers-30pct-2030/
URL:https://sipoly.si/claim/leap-ai-engaged-papers-30pct-2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-415
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20311231
DTEND;VALUE=DATE:20320101
SUMMARY:Due: AI will use 7% of US electricity in 2030 (LEAP expert panel)
DESCRIPTION:“That rises to 7% of all electricity consumption in 2030” (
 LEAP expert panel\, said Nov 10\, 2025)\n\nHow we judge it: Our interpreta
 tion: correct if credible 2030 estimates put AI training plus inference at
  5.5-8.5% of US electricity consumption.\n\nScorecard: https://sipoly.si/c
 laim/leap-ai-electricity-7pct-2030/
URL:https://sipoly.si/claim/leap-ai-electricity-7pct-2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-424
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20311231
DTEND;VALUE=DATE:20320101
SUMMARY:Due: Data centre electricity demand will more than double to ~945 T
 Wh by 2030 (International Energy Agency)
DESCRIPTION:“It projects that electricity demand from data centres worldw
 ide is set to more than double by 2030 to around 945 terawatt-hours (TWh)\
 , slightly more than the entire electricity consumption of Japan today.”
  (International Energy Agency\, said Apr 10\, 2025)\n\nHow we judge it: Ou
 r interpretation: correct if IEA (or comparable) data put global data-cent
 re electricity consumption in 2030 at 850-1\,050 TWh.\n\nScorecard: https:
 //sipoly.si/claim/iea-datacentre-945twh-2030/
URL:https://sipoly.si/claim/iea-datacentre-945twh-2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-313
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20311231
DTEND;VALUE=DATE:20320101
SUMMARY:Due: A major city bans human-driven cars from part of town for driv
 erless cars: 2027 to 2031 (Rodney Brooks)
DESCRIPTION:“A major city bans parking and cars with drivers from a non-t
 rivial portion of a city so that driverless cars have free reign in that a
 rea. NET 2027 BY 2031” (Rodney Brooks\, said Jan 1\, 2018)\n\nHow we jud
 ge it: Our interpretation: correct if a major city first bans parking and 
 human-driven cars from a non-trivial area in favour of driverless cars in 
 2027-2031 (not before 2027\, and by end of 2031).\n\nScorecard: https://si
 poly.si/claim/brooks-city-bans-human-drivers-net2027/
URL:https://sipoly.si/claim/brooks-city-bans-human-drivers-net2027/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-437
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20311231
DTEND;VALUE=DATE:20320101
SUMMARY:Due: AI will outperform humans at working in retail by 2031 (median
 ) (ML researchers (2016 expert survey))
DESCRIPTION:“Researchers predict AI will outperform humans in many activi
 ties in the next ten years\, such as translating languages (by 2024)\, wri
 ting high-school essays (by 2026)\, driving a truck (by 2027)\, working in
  retail (by 2031)\, writing a bestselling book (by 2049)\, and working as 
 a surgeon (by 2053).” (ML researchers (2016 expert survey)\, said May 24
 \, 2017)\n\nHow we judge it: Our interpretation: by end-2031\, AI/robotic 
 systems can do the full job of a retail salesperson at least as well as hu
 mans (deployed beyond pilots).\n\nScorecard: https://sipoly.si/claim/ml-su
 rvey-2016-retail-2031/
URL:https://sipoly.si/claim/ml-survey-2016-retail-2031/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-362
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20321231
DTEND;VALUE=DATE:20330101
SUMMARY:Due: 50/50 that AI learns on the job like a human for any white-col
 lar work by 2032 (Dwarkesh Patel)
DESCRIPTION:“AI learns on the job as easily\, organically\, seamlessly\, 
 and quickly as a human\, for any white collar work.” (Dwarkesh Patel\, s
 aid Jun 2\, 2025)\n\nHow we judge it: Our interpretation: YES if by 31 Dec
  2032 deployed AI systems show human-like continual learning on the job ac
 ross white-collar work.\n\nScorecard: https://sipoly.si/claim/dwarkesh-on-
 the-job-learning-2032/
URL:https://sipoly.si/claim/dwarkesh-on-the-job-learning-2032/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-409
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20330101
DTEND;VALUE=DATE:20330102
SUMMARY:Due: 45%: AI capable of fully automating AI R&D by the start of 203
 3 (Ryan Greenblatt)
DESCRIPTION:“I now would put ~15% probability by the start of 2029 and a 
 45% chance by the start of 2033.” (Ryan Greenblatt\, said Aug 2025)\n\nH
 ow we judge it: Our interpretation: YES if by 1 Jan 2033 an AI system exis
 ts that could fully automate AI R&D.\n\nScorecard: https://sipoly.si/claim
 /greenblatt-ai-rd-automation-2033-45pct/
URL:https://sipoly.si/claim/greenblatt-ai-rd-automation-2033-45pct/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-355
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20331231
DTEND;VALUE=DATE:20340101
SUMMARY:Due: AGI "slightly after" 2030\, not quite by 2030 (Sundar Pichai)
DESCRIPTION:“I don’t think we’ll quite get there by 2030\, so my sens
 e is it’s slightly after that” (Sundar Pichai\, said Jun 5\, 2025)\n\n
 How we judge it: Our interpretation: correct if AGI (as broadly recognized
 ) has not arrived by the end of 2030 but does arrive by about 2033.\n\nSco
 recard: https://sipoly.si/claim/pichai-agi-slightly-after-2030/
URL:https://sipoly.si/claim/pichai-agi-slightly-after-2030/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-345
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20331231
DTEND;VALUE=DATE:20340101
SUMMARY:Due: Generative AI could raise global GDP by 7% over ten years (Gol
 dman Sachs Research)
DESCRIPTION:“they could drive a 7% (or almost $7 trillion) increase in gl
 obal GDP and lift productivity growth by 1.5 percentage points over a 10-y
 ear period.” (Goldman Sachs Research\, said Apr 5\, 2023)\n\nHow we judg
 e it: Our interpretation: by 2033\, US/global productivity growth runs abo
 ut 1.5 percentage points above its pre-2023 trend\, as estimated by mainst
 ream sources. Hedged ("could").\n\nScorecard: https://sipoly.si/claim/gold
 man-genai-gdp-7pct-10y/
URL:https://sipoly.si/claim/goldman-genai-gdp-7pct-10y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-403
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20340113
DTEND;VALUE=DATE:20340114
SUMMARY:Due: AI will displace about 40% of the world's jobs within 15 years
  (Kai-Fu Lee)
DESCRIPTION:“all together in 15 years\, that's going to displace about 40
 % of the jobs in the world.” (Kai-Fu Lee\, said Jan 13\, 2019)\n\nHow we
  judge it: Our interpretation: by Jan 2034\, roughly 40% of jobs worldwide
  have been displaced by AI and automation (eliminated or replaced\, not ju
 st changed).\n\nScorecard: https://sipoly.si/claim/kai-fu-lee-40pct-jobs-1
 5y/
URL:https://sipoly.si/claim/kai-fu-lee-40pct-jobs-15y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-368
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20340621
DTEND;VALUE=DATE:20340622
SUMMARY:Due: AI "10\,000 times smarter than humans" within 10 years (Masayo
 shi Son)
DESCRIPTION:“Artificial intelligence that is 10\,000 times smarter than h
 umans will be here in 10 years\, SoftBank CEO Masayoshi Son said on Friday
  in a rare public appearance.” (Masayoshi Son\, said Jun 21\, 2024)\n\nH
 ow we judge it: Our interpretation: by 21 Jun 2034\, artificial superintel
 ligence vastly exceeding the best humans at essentially all cognitive task
 s exists. "10\,000 times" is not measurable\, so we score the ASI part.\n\
 nScorecard: https://sipoly.si/claim/son-asi-10000x-10-years/
URL:https://sipoly.si/claim/son-asi-10000x-10-years/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-683
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20341231
DTEND;VALUE=DATE:20350101
SUMMARY:Due: AI will add less than 0.55% to total factor productivity over 
 10 years (Daron Acemoglu)
DESCRIPTION:“Consequently\, predicted TFP gains over the next 10 years ar
 e even more modest and are predicted to be less than 0.55%.” (Daron Acem
 oglu\, said Apr 5\, 2024)\n\nHow we judge it: Our interpretation: by 2034\
 , mainstream estimates (e.g. BLS/Fed/OECD decompositions) attribute less t
 han 0.55% cumulative US TFP gain to AI over 2024-2034.\n\nScorecard: https
 ://sipoly.si/claim/acemoglu-tfp-under-055-10y/
URL:https://sipoly.si/claim/acemoglu-tfp-under-055-10y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-308
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20341231
DTEND;VALUE=DATE:20350101
SUMMARY:Due: Affordable home-navigating robot: not before 2035 (Rodney Broo
 ks)
DESCRIPTION:“Expensive product: NET 2030 Affordable product: NET 2035 Wha
 t is easy for humans is still very\, very hard for robots.” (Rodney Broo
 ks\, said Jan 1\, 2018)\n\nHow we judge it: Our interpretation: correct if
  no affordable (mass-market consumer price) robot able to navigate almost 
 any US home is on sale before 1 Jan 2035.\n\nScorecard: https://sipoly.si/
 claim/brooks-home-robot-affordable-net2035/
URL:https://sipoly.si/claim/brooks-home-robot-affordable-net2035/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-353
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20350228
DTEND;VALUE=DATE:20350301
SUMMARY:Due: Within a decade AI makes great medical advice and tutoring "fr
 ee\, commonplace" (Bill Gates)
DESCRIPTION:“with AI\, over the next decade\, that will become free\, com
 monplace — great medical advice\, great tutoring” (Bill Gates\, said F
 eb 2025)\n\nHow we judge it: Our interpretation: by Feb 2035\, AI medical 
 advice and tutoring of a quality comparable to a great doctor or teacher a
 re freely and widely available.\n\nScorecard: https://sipoly.si/claim/gate
 s-humans-not-needed-decade/
URL:https://sipoly.si/claim/gates-humans-not-needed-decade/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-359
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20350319
DTEND;VALUE=DATE:20350320
SUMMARY:Due: Within a decade\, AI agents will independently do many softwar
 e tasks that take humans days or weeks (METR)
DESCRIPTION:“Extrapolating this trend predicts that\, in under a decade\,
  we will see AI agents that can independently complete a large fraction of
  software tasks that currently take humans days or weeks.” (METR\, said 
 Mar 19\, 2025)\n\nHow we judge it: Our interpretation: by 19 Mar 2035\, ME
 TR (or equivalent) reports frontier agents with a 50% time horizon of at l
 east one work-week (40 hours) on software tasks.\n\nScorecard: https://sip
 oly.si/claim/metr-week-long-tasks-decade/
URL:https://sipoly.si/claim/metr-week-long-tasks-decade/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-682
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20350415
DTEND;VALUE=DATE:20350416
SUMMARY:Due: Transformative economic and societal impacts of AI will take d
 ecades (Arvind Narayanan)
DESCRIPTION:“we think that transformative economic and societal impacts w
 ill be slow (on the timescale of decades)” (Arvind Narayanan\, said Apr 
 15\, 2025)\n\nHow we judge it: Our interpretation: correct if by 15 Apr 20
 35 there has been no economy-wide AI transformation (e.g. no decade of US 
 labor-productivity growth above 3%/yr\, no AI-driven unemployment shock ab
 ove 10%).\n\nScorecard: https://sipoly.si/claim/narayanan-kapoor-impacts-d
 ecades/
URL:https://sipoly.si/claim/narayanan-kapoor-impacts-decades/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-360
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20350430
DTEND;VALUE=DATE:20350501
SUMMARY:Due: Only about 30% chance of full remote-work automation within 10
  years (Ege Erdil)
DESCRIPTION:“I still think full automation of remote work in 10 years is 
 plausible\, because it’s what we would predict if we straightforwardly e
 xtrapolate current rates of revenue growth and assume no slowdown. However
 \, I would only give this outcome around 30% chance.” (Ege Erdil\, said 
 Apr 2025)\n\nHow we judge it: Our interpretation: YES if\, by Apr 2035\, A
 I can do essentially all remote (computer-based) jobs at or below human co
 st.\n\nScorecard: https://sipoly.si/claim/erdil-remote-work-automation-10y
 -30pct/
URL:https://sipoly.si/claim/erdil-remote-work-automation-10y-30pct/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-78
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20350604
DTEND;VALUE=DATE:20350605
SUMMARY:Due: Maybe 50% chance of AGI in five to ten years (Demis Hassabis)
DESCRIPTION:“In the next five to 10 years\, there’s maybe a 50 percent 
 chance that we'll have what we define as AGI.” (Demis Hassabis\, said Ju
 n 4\, 2025)\n\nHow we judge it: Our interpretation: by 4 Jun 2035\, a syst
 em exhibits all the cognitive capabilities humans have\, per broad expert 
 consensus.\n\nScorecard: https://sipoly.si/claim/hassabis-agi-50pct-5-10y/
URL:https://sipoly.si/claim/hassabis-agi-50pct-5-10y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-344
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20351017
DTEND;VALUE=DATE:20351018
SUMMARY:Due: Truly capable AI agents are about "a decade" away\, not "the y
 ear of agents" (Andrej Karpathy)
DESCRIPTION:“I feel like the problems are tractable\, they’re surmounta
 ble\, but they’re still difficult. If I just average it out\, it just fe
 els like a decade to me.” (Andrej Karpathy\, said Oct 17\, 2025)\n\nHow 
 we judge it: Our interpretation: "about a decade" means agents that can be
  hired like a reliable employee or intern for general knowledge work do no
 t arrive before about Oct 2030 and do arrive by about Oct 2035. Scored at 
 the end of the window.\n\nScorecard: https://sipoly.si/claim/karpathy-agen
 ts-decade/
URL:https://sipoly.si/claim/karpathy-agents-decade/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-322
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20351112
DTEND;VALUE=DATE:20351113
SUMMARY:Due: Strong AGI expected "in the next 5-10 years" (from Nov 2025) (
 Daniel Kokotajlo)
DESCRIPTION:“The companies seem to think strong AGI is just a few years a
 way\, and while I'm not as bullish as they are\, I do expect it to happen 
 in the next 5-10 years.” (Daniel Kokotajlo\, said Nov 12\, 2025)\n\nHow 
 we judge it: Our interpretation: by 12 Nov 2035\, an AI system broadly acc
 epted as strong AGI (able to do essentially all cognitive work humans do) 
 exists.\n\nScorecard: https://sipoly.si/claim/kokotajlo-strong-agi-5-10y/
URL:https://sipoly.si/claim/kokotajlo-strong-agi-5-10y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-667
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20360101
DTEND;VALUE=DATE:20360102
SUMMARY:Due: Humanoid dexterity "will remain pathetic" compared to human ha
 nds through 2035 (Rodney Brooks)
DESCRIPTION:“Humanoid Robots . Deployable dexterity will remain pathetic 
 compared to human hands beyond 2036. Without new types of mechanical syste
 ms walking humanoids will remain too unsafe to be in close proximity to re
 al humans.” (Rodney Brooks\, said Mar 26\, 2025)\n\nHow we judge it: Our
  interpretation: correct if on 1 Jan 2036 no deployed humanoid robot has h
 and dexterity broadly comparable to human hands\, and walking humanoids ar
 e not working in close proximity to people at scale.\n\nScorecard: https:/
 /sipoly.si/claim/brooks-2025-humanoid-dexterity/
URL:https://sipoly.si/claim/brooks-2025-humanoid-dexterity/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-314
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20360101
DTEND;VALUE=DATE:20360102
SUMMARY:Due: Only Waymo and Zoox will matter for US self-driving through 20
 35 (Rodney Brooks)
DESCRIPTION:“In the US the players that will determine whether self drivi
 ng cars are successful or abandoned are #1 Waymo (Google) and #2 Zoox (Ama
 zon). No one else matters.” (Rodney Brooks\, said Mar 26\, 2025)\n\nHow 
 we judge it: Our interpretation: correct if at 1 Jan 2036 Waymo and Zoox a
 re the two largest US driverless ride services by rides or fleet\, and no 
 other company (e.g. Tesla) runs a comparable driverless service.\n\nScorec
 ard: https://sipoly.si/claim/brooks-2025-waymo-zoox-only/
URL:https://sipoly.si/claim/brooks-2025-waymo-zoox-only/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-689
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20361231
DTEND;VALUE=DATE:20370101
SUMMARY:Due: ~35% probability of transformative AI by 2036 (Ajeya Cotra)
DESCRIPTION:“~35% probability by 2036 (a ~3x likelihood ratio [3] vs 15%)
 .” (Ajeya Cotra\, said Aug 2\, 2022)\n\nHow we judge it: Our interpretat
 ion: YES if transformative AI (as defined in her report) exists by 31 Dec 
 2036.\n\nScorecard: https://sipoly.si/claim/cotra-tai-2036-35pct/
URL:https://sipoly.si/claim/cotra-tai-2036-35pct/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-442
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20371231
DTEND;VALUE=DATE:20380101
SUMMARY:Due: 10% chance all human occupations are fully automatable by 2037
  (50% by 2116) (AI researchers (2023 expert survey))
DESCRIPTION:“However\, the chance of all human occupations becoming fully
  automatable was forecast to reach 10% by 2037\, and 50% as late as 2116
 ” (AI researchers (2023 expert survey)\, said Jan 5\, 2024)\n\nHow we ju
 dge it: Our interpretation: YES if by end-2037 every human occupation coul
 d be fully automated by AI at lower cost than human workers.\n\nScorecard:
  https://sipoly.si/claim/ai-survey-2023-full-automation-10pct-2037/
URL:https://sipoly.si/claim/ai-survey-2023-full-automation-10pct-2037/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-690
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20401231
DTEND;VALUE=DATE:20410101
SUMMARY:Due: Median ~2040 for transformative AI (50% by 2040) (Ajeya Cotra)
DESCRIPTION:“A median of ~2040 (a decrease of ~10 years from 2050).” (A
 jeya Cotra\, said Aug 2\, 2022)\n\nHow we judge it: Our interpretation: YE
 S if transformative AI (as defined in her report) exists by 31 Dec 2040.\n
 \nScorecard: https://sipoly.si/claim/cotra-tai-median-2040/
URL:https://sipoly.si/claim/cotra-tai-median-2040/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-423
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20431231
DTEND;VALUE=DATE:20440101
SUMMARY:Due: Transformative AGI by 2043 is less than 1% likely (Ted Sanders
 )
DESCRIPTION:“we estimate the likelihood of transformative artificial gene
 ral intelligence (AGI) by 2043 and find it to be <1%.” (Ted Sanders\, sa
 id Jun 6\, 2023)\n\nHow we judge it: Our interpretation: YES if by end-204
 3 AI can perform nearly all economically valuable tasks at or below human 
 cost (the contest definition).\n\nScorecard: https://sipoly.si/claim/sande
 rs-transformative-agi-2043-under-1pct/
URL:https://sipoly.si/claim/sanders-transformative-agi-2043-under-1pct/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-352
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20441227
DTEND;VALUE=DATE:20441228
SUMMARY:Due: AIs smarter than people "within probably the next 20 years" (G
 eoffrey Hinton)
DESCRIPTION:“most of the experts in the field think that sometime\, withi
 n probably the next 20 years\, we’re going to develop AIs that are smart
 er than people.” (Geoffrey Hinton\, said Dec 27\, 2024)\n\nHow we judge 
 it: Our interpretation: by 27 Dec 2044\, AI systems exceed human experts a
 t most cognitive tasks.\n\nScorecard: https://sipoly.si/claim/hinton-smart
 er-than-people-20y/
URL:https://sipoly.si/claim/hinton-smarter-than-people-20y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-356
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20451125
DTEND;VALUE=DATE:20451126
SUMMARY:Due: A system that learns as well as a human\, then becomes superhu
 man: "5 to 20" years (Ilya Sutskever)
DESCRIPTION:“I think like 5 to 20.” (Ilya Sutskever\, said Nov 25\, 202
 5)\n\nHow we judge it: Our interpretation: correct if such a system arrive
 s between Nov 2030 and Nov 2045\; wrong if before Nov 2030 or not by Nov 2
 045.\n\nScorecard: https://sipoly.si/claim/sutskever-superhuman-learner-5-
 20y/
URL:https://sipoly.si/claim/sutskever-superhuman-learner-5-20y/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-704
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20451231
DTEND;VALUE=DATE:20460101
SUMMARY:Due: "We are going to expand intelligence a millionfold by 2045" (R
 ay Kurzweil)
DESCRIPTION:“We are going to expand intelligence a millionfold by 2045 an
 d it is going to deepen our awareness and consciousness.” (Ray Kurzweil\
 , said Jun 29\, 2024)\n\nHow we judge it: Our interpretation: by 2045\, hu
 mans routinely augment their intelligence through brain-computer interface
 s connected to AI\, and effective (human plus machine) intelligence is man
 y orders of magnitude above 2024 levels.\n\nScorecard: https://sipoly.si/c
 laim/kurzweil-millionfold-2045/
URL:https://sipoly.si/claim/kurzweil-millionfold-2045/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-441
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20471231
DTEND;VALUE=DATE:20480101
SUMMARY:Due: 50% chance machines outperform humans at every task by 2047 (A
 I researchers (2023 expert survey))
DESCRIPTION:“If science continues undisrupted\, the chance of unaided mac
 hines outperforming humans in every possible task was estimated at 10% by 
 2027\, and 50% by 2047.” (AI researchers (2023 expert survey)\, said Jan
  5\, 2024)\n\nHow we judge it: Our interpretation: YES if by end-2047 unai
 ded AI systems can do every task better and more cheaply than human worker
 s.\n\nScorecard: https://sipoly.si/claim/ai-survey-2023-all-tasks-50pct-20
 47/
URL:https://sipoly.si/claim/ai-survey-2023-all-tasks-50pct-2047/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-666
DTSTAMP:20261009T000000Z
DTSTART;VALUE=DATE:20471231
DTEND;VALUE=DATE:20480101
SUMMARY:Due: A robot as intelligent\, attentive and faithful as a dog: not 
 before 2048 (Rodney Brooks)
DESCRIPTION:“A robot that seems as intelligent\, as attentive\, and as fa
 ithful\, as a dog. NET 2048” (Rodney Brooks\, said Jan 1\, 2018)\n\nHow 
 we judge it: Our interpretation: correct if no robot is broadly judged as 
 intelligent\, attentive and faithful as a dog before 1 Jan 2048.\n\nScorec
 ard: https://sipoly.si/claim/brooks-dog-level-robot-net2048/
URL:https://sipoly.si/claim/brooks-dog-level-robot-net2048/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-438
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20491231
DTEND;VALUE=DATE:20500101
SUMMARY:Due: AI will write a bestselling book by 2049 (median) (ML research
 ers (2016 expert survey))
DESCRIPTION:“Researchers predict AI will outperform humans in many activi
 ties in the next ten years\, such as translating languages (by 2024)\, wri
 ting high-school essays (by 2026)\, driving a truck (by 2027)\, working in
  retail (by 2031)\, writing a bestselling book (by 2049)\, and working as 
 a surgeon (by 2053).” (ML researchers (2016 expert survey)\, said May 24
 \, 2017)\n\nHow we judge it: Our interpretation: by end-2049\, an AI-writt
 en book (credited as such) reaches the New York Times bestseller list.\n\n
 Scorecard: https://sipoly.si/claim/ml-survey-2016-bestseller-2049/
URL:https://sipoly.si/claim/ml-survey-2016-bestseller-2049/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-439
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20531231
DTEND;VALUE=DATE:20540101
SUMMARY:Due: AI will outperform humans at working as a surgeon by 2053 (med
 ian) (ML researchers (2016 expert survey))
DESCRIPTION:“Researchers predict AI will outperform humans in many activi
 ties in the next ten years\, such as translating languages (by 2024)\, wri
 ting high-school essays (by 2026)\, driving a truck (by 2027)\, working in
  retail (by 2031)\, writing a bestselling book (by 2049)\, and working as 
 a surgeon (by 2053).” (ML researchers (2016 expert survey)\, said May 24
 \, 2017)\n\nHow we judge it: Our interpretation: by end-2053\, autonomous 
 surgical systems perform at least as well as human surgeons across common 
 operations.\n\nScorecard: https://sipoly.si/claim/ml-survey-2016-surgeon-2
 053/
URL:https://sipoly.si/claim/ml-survey-2016-surgeon-2053/
TRANSP:TRANSPARENT
END:VEVENT
BEGIN:VEVENT
UID:sipoly.si-claim-351
DTSTAMP:20261008T000000Z
DTSTART;VALUE=DATE:20541227
DTEND;VALUE=DATE:20541228
SUMMARY:Due: "10% to 20%" chance AI leads to human extinction within three 
 decades (Geoffrey Hinton)
DESCRIPTION:“said there was a “10% to 20%” chance that AI would lead 
 to human extinction within the next three decades.” (Geoffrey Hinton\, s
 aid Dec 27\, 2024)\n\nHow we judge it: Our interpretation: a probability s
 tatement over 30 years\; it cannot be scored on a single outcome before 20
 54. Kept for the record.\n\nScorecard: https://sipoly.si/claim/hinton-exti
 nction-10-20pct-30y/
URL:https://sipoly.si/claim/hinton-extinction-10-20pct-30y/
TRANSP:TRANSPARENT
END:VEVENT
END:VCALENDAR