A summary of AI surveys

By Katja Grace, 10 January 2015

If you want to know when human-level AI will be developed, a natural approach is to ask someone who works on developing AI. You might however be put off by such predictions being regularly criticized as inaccurate and biased. While they do seem overwhelmingly likely to be inaccurate and biased, I claim they would have to be very inaccurate and biased before they were worth ignoring, especially in the absence of many other sources of quality information. The bar for ridicule is well before the bar for being uninformative.

So on that note, we made a big summary of all of the surveys we know of on timelines to human-level AI. And also a bunch of summary pages on specific human-level AI surveys. We hope they are a useful reference, and also help avert selection bias selection bias from people only knowing about surveys that support their particular views on selection bias.

It’s interesting to note the consistency between the surveys that asked participants to place confidence intervals. They all predict there is a ten percent chance of human-level AI sometime in the 2020s, and almost all place a fifty percent chance of human-level AI between 2040 and 2050. They are even pretty consistent on the 90% date, with more than half in 2070-2080. This is probably mostly evidence that people talk to each other and hear about similar famous predictions. However it is some evidence of accuracy, since if each survey produced radically different estimates we must conclude that surveys are fairly inaccurate.

If you know of more surveys on human-level AI timelines, do send them our way.

Here’s a summary of our summary:

Year Survey #  10%  50%  90%  Other key ‘Predictions’ Participants Response rate Link to original document
1972  Michie 67 Median 50y (2022) (vs 20 or >50) AI, CS link
2005  Bainbridge 26  Median 2085 Tech  link
2006  AI@50 median >50y (2056) AI conf link
2007  Klein 888 median 2030-2050 Futurism? link and link
2009  AGI-09  2020  2040  2075 AGI conf; AI link
2011  FHI Winter Intelligence 35  2028 2050  2150 AGI impacts conf; 44% related technical 41% link
2011-2012  Kruel interviews 37  2025  2035  2070 AGI, AI link
2012  FHI: AGI 72  2022  2040  2065 AGI & AGI impacts conf; AGI, technical work 65% link
2012  FHI:PT-AI 43  2023  2048  2080 Philosophy & theory of AI conf; not technical AI 49% link
2012-present  Hanson ~10  ≤ 10% progress to human level in past 20y AI link
2013  FHI: TOP100 29 2022  2040  2075 Top AI 29% link
2013  FHI:EETN 26  2020  2050  2093 Greek assoc. for AI; AI 10% link

 

(Image: AGI-09 participants, by jeriaska)


We welcome suggestions for this page or anything on the site via our feedback box, though will not address all of them.

2 Trackbacks / Pingbacks

  1. Day 2: Shortest paths, Arrogance – Ben Models the World
  2. Day 2: Shortest paths, Arrogance « Ben Models the World

Leave a Reply

Your email address will not be published.


*


This site uses Akismet to reduce spam. Learn how your comment data is processed.