Among the forecasts he was describing, which averaged a hundred years, nobody said it would happen within 30 years, but now everybody is saying that.

Judge as of 2024-07-09. Two parts: (a) when Kurzweil made his 1999 prediction (human-level AI by 2029), nobody else predicted AGI within 30 years; (b) by 2024 'everybody' (read as experts and leading AI figures broadly) says AGI will arrive within about 30 years or sooner. True: the record shows essentially no other notable forecasters predicting AGI within about 30 years in 1999, and by 2024 a clear majority of experts and leading AI figures expected AGI within that window. Mostly true: in 1999 such predictions were rare and on the fringe (a few exceptions, e.g. Moravec), and by 2024 most prominent figures and surveyed experts held near-term timelines. Mixed: one part holds and the other fails, e.g. 2024 expert surveys show the median far from consensus, or several notable people in 1999 predicted similar timelines. Mostly false: there is a real shift toward shorter timelines, but both parts are greatly overstated (others such as Moravec or Vinge did predict it within 30 years, and 2024 surveys show widespread disagreement or median dates after 2054). False: the evidence shows many contemporaries predicted it within 30 years in 1999, and/or 2024 expert opinion did not mostly expect AGI within 30 years.

Set before the research began.

1. E1 The Singularity and Superintelligence Geschichte der Informatik

   “Its abstract opened: “Within thirty years, we will have the technological means to create superhuman intelligence. Shortly after, the human era will be ended.””

2. E2 The Singularity and Superintelligence Geschichte der Informatik · 1988

   “The roboticist Hans Moravec, in Mind Children (1988), extrapolated computing power to estimate that supercomputers would match the human brain before 2010.”

3. E3 AGI by 2029: The Turing Threshold askhistoryai.com

   “by Kurzweil's account, a circa-2000 conference of AI experts broadly expected human-level AI to take closer to a century rather than three decades”

4. E4 2023 Expert Survey on Progress in AI [AI Impacts Wiki] AI Impacts · 17 August 2023

   “The 2023 Expert Survey on Progress in AI is a survey of 2,778 AI researchers that AI Impacts ran in October 2023.”

5. E5 2023 Expert Survey on Progress in AI [AI Impacts Wiki] AI Impacts · 17 August 2023

   “The resulting aggregate forecast gives a 50% chance of HLMI by 2047, down thirteen years from 2060 in the 2022 ESPAI.”

6. E6 2023 Expert Survey on Progress in AI [AI Impacts Wiki] AI Impacts · 17 August 2023

   “The 774 responses to this question were used to create an aggregate forecast which gives a 50% chance of FAOL by 2116, down 48 years from 2164 in the 2022 ESPAI.”

7. E7 AGI by 2029: The Turing Threshold askhistoryai.com (unattributed site)

   “Kurzweil first published it in 1999's The Age of Spiritual Machines, when a Stanford conference of AI experts agreed his scenario was plausible but insisted it would take a century, not thirty years.”

8. E8 AGI by 2029: The Turing Threshold askhistoryai.com (unattributed site)

   “In the past I've disagreed about how soon super intelligence was coming, and now I think we're pretty much agreed.”

9. E9 Thousands of AI Authors on the Future of AI AI Impacts (Katja Grace, Harlan Stewart, Julia Fabienne Sandkühler, Stephen Thomas, Ben Weinstein-Raun, Jan Brauner) · January 2024

   “The aggregate 2023 forecast predicted a 50% chance of HLMI by 2047, down thirteen years from 2060 in the 2022 survey.”

10. E10 Thousands of AI Authors on the Future of AI AI Impacts (Katja Grace, Harlan Stewart, Julia Fabienne Sandkühler, Stephen Thomas, Ben Weinstein-Raun, Jan Brauner) · January 2024

    “While the range of views on how long it will take for milestones to be feasible can be broad, this year’s survey saw a general shift towards earlier expectations.”

11. E11 Reason.tv: A Conversation with Author Vernor Vinge Reason Magazine (Reason Foundation) · 29 August 2011

    “In Vinge's 1993 essay "The Coming Technological Singularity" Vinge wrote, "Within thirty years, we will have the technological means to create superhuman intelligence. Shortly after, the human era will be ended."”

On part (a), the evidence shows notable forecasters before 1999 with similar or shorter timelines: Vinge wrote in 1993 that superhuman intelligence would arrive "within thirty years" (E11, E1), and Moravec in 1988 projected that supercomputers would match the human brain before 2010 (E2). Reports that experts around 2000 expected AI to take about a century rest on Kurzweil's own account (E3, E7). On part (b), the 2023 AI Impacts survey's aggregate forecast gave a 50% chance of human-level machine intelligence by 2047, which falls within 30 years of 2024 and is thirteen years earlier than the previous survey (E5, E9). However, 50% by 2047 does not amount to "everybody": the survey reports a broad range of views (E10), and its forecast for full automation of labor is 2116 (E6). So the claim of a shift toward shorter timelines holds up, but both parts are overstated, and the result is mixed.