2027AI company · said Mar 6, 2025
About: powerful AI: “Intellectual capabilities matching or exceeding that of Nobel Prize winners across most disciplines”
“we expect powerful AI systems will emerge in late 2026 or early 2027.”
When: late 2026 or early 2027
Anthropic's recommendations to the White House Office of Science and Technology Policy for the US AI Action Plan, Mar 6, 2025 · company-reported
The full record2028Co-founder and Chief AGI Scientist, Google DeepMind · said Oct 26, 2023
About: AGI
“And so, yeah, I think there's a 50% chance that we have AGI by 2028.”
When: 2028 (50% chance)
Dwarkesh Podcast, Oct 26, 2023
The full record2028AI Futures Project · said Apr 2, 2026
About: “Automated Coder”: AI that can do the coding work of an AI company
“Daniel's Automated Coder (AC) median has moved from late 2029 to mid 2028.”
When: mid 2028 (Kokotajlo median); mid 2030 (Lifland median, moved from early 2032)
AI Futures Project, “Q1 2026 Timelines Update”, Apr 2, 2026
The full record2028Executive Director, AI Futures Project; co-author of AI 2027 · said Sep 30, 2026
About: AI that automates AI research
“I personally would guess about 50% chance by end of 2028.”
Also said: “It's unclear when they will succeed but my team and I think it could happen any year now.”
When: end of 2028 (50%)
Testimony to the US Senate, posted on the AI Futures Project blog, Oct 2, 2026
The full record2029CEO, Nvidia · said Mar 1, 2024
About: AI passing “every single test that you can possibly imagine”, his answer to a question on how long AGI would take
“and I'm guessing in five years time, we'll do well on every single one.”
Also said: “He said that by other definitions AGI “may be much further away” (Reuters' paraphrase with quoted words).”
When: about 2029
Economic forum at Stanford University (Reuters report, 3 Mar 2024), Mar 3, 2024
The full record2029Futurist; Principal Researcher, Google · said Jun 14, 2024
About: AGI and the Turing test
“In 1999, I predicted by 2029, we'd reach the Turing test and also AGI, where basically a computer can do anything a human can do.”
Also said: “Asked if his 2045 Singularity date stands: “I'm sticking with that prediction.””
When: 2029
Science Friday (radio), on his book The Singularity Is Nearer, Jun 14, 2024
The full record2030CEO, Google DeepMind · said Jan 23, 2025
About: AGI
“I think we're probably three to five years away.”
When: 2028–2030 (counted from Jan 2025)
Big Technology (Alex Kantrowitz) interview, Jan 23, 2025
The full record2030CEO, Zhipu AI · said Sep 30, 2025
About: artificial superintelligence by 2030 (headline: full ASI unlikely by 2030)
“I think achieving or exceeding human intelligence levels by 2030 might mean surpassing humans in one or several aspects”
When: 2030: partial only
At the launch of Zhipu's GLM-4.6 model, Beijing (Reuters), Sep 30, 2025
The full record2033OpenAI CEO, president and then chief scientist · said May 22, 2023
About: AI exceeding expert skill (in a post on superintelligence)
“it's conceivable that within the next ten years, AI systems will exceed expert skill level in most domains”
When: by about 2033
OpenAI blog post “Governance of superintelligence”, May 22, 2023 · company-reported
The full record2034CEO, SoftBank · said Jun 21, 2024
About: artificial superintelligence; CNBC reports he said AI “10,000 times smarter than humans will be here in 10 years”
“But when it comes to ASI it's a totally different story. [With] ASI, you will see a big improvement.”
When: ASI by about 2034; AGI in three to five years (CNBC's summary)
SoftBank annual shareholder meeting, live translation (reported by CNBC, Arjun Kharpal), Jun 21, 2024
The full record2047AI Impacts; survey of authors who published at top AI venues · said Jan 5, 2024
About: high-level machine intelligence
“the chance of unaided machines outperforming humans in every possible task was estimated at 10% by 2027, and 50% by 2047.”
Also said: “the chance of all human occupations becoming fully automatable was forecast to reach 10% by 2037, and 50% as late as 2116”
When: 2047 (50%)
“Thousands of AI Authors on the Future of AI” (arXiv; later in JAIR), Jan 5, 2024
The full record?CEO, Baidu · said May 2024
About: AGI
“AGI is still quite a few years away.”
Also said: “I think [it] is more than 10 years away.”
When: more than 10 years (after 2034)
VivaTech, Paris (reported by CNBC, Arjun Kharpal), May 23, 2024
The full record?Founder and CEO, DeepSeek · said Jul 2024
About: AGI
“It could be two, five, or ten years–in any case, it will happen in our lifetimes.”
Also said: “Our ultimate goal is still to achieve AGI”
When: 2026–2034 (two to ten years from July 2024)
Interview with 36Kr's “Waves” (暗涌), English translation by ChinaTalk, Nov 27, 2024
The full record?CEO, OpenAI · said Sep 23, 2024
About: superintelligence
“It is possible that we will have superintelligence in a few thousand days (!)”
When: 'a few thousand days' from Sept 2024 (vague; 2,000–9,000 days would be roughly 2030–2049)
Essay “The Intelligence Age”, Sep 23, 2024 · company-reported
The full record?AI researcher, University of Toronto (often called a “godfather of AI”) · said Jan 2025
About: superintelligence
“Now I think it's maybe five to 20 years before we get superintelligence.”
Also said: “I thought it would be 50 to 100 years.” “Maybe longer, but it's coming quicker than I thought.”
When: about 2030–2045 (counted from Jan 2025)
CBC News interview (Lauren Sproule), Jan 11, 2025
The full record?Cognitive scientist and AI critic, NYU professor emeritus · said Nov 21, 2025
About: AGI
“AGI ain't actually coming in 2027”
When: not in 2027; also wrote AGI “is NOT in fact coming in Trump's second term, and never was” (i.e. not before Jan 2029)
The bet: Bet with Miles Brundage (30 Dec 2024): if AI systems can do 8 of 10 named tasks “by the end of 2027” Marcus donates $2,000; if fewer than 8, Brundage donates $20,000. Terms, Dec 30, 2024
Marcus on AI (Substack), Nov 21, 2025 · company-reported
The full record