This AI Prediction was made by all researchers who published at the 2015 NIPS and ICML conferences[…]A total of 352 researchers responded to our survey invitation (21% of the 1634 authors we contacted) in 2016.
Predicted time for AGI / HLMI / transformative AI:
(Hover for explanation)Types of advanced artificial intelligence: AGI (AI that can perform many tasks at a human-level), HLMI (more advanced AI that surpasses human intelligence in specific areas), and Transformative AI (AI that could significantly impact society and the world)
[Researchers believe there is a 50% chance of AI outperforming humans in all tasks in 45 years and of automating all human jobs in 120 yearsMedian dates and probabilities under the fixed probabilitiesHLMI10%: 15 years50%: 40 years90%: 100 yearsFull Automation of labor10%: 50 years50%: 90 years90%: 200 yearsMedian dates and probabilities under the fixed yearsHLMI10 years: 1 years: 10 years: 30%Full Automation of labor10 years: 0 years: 0.01 years: 3%]
Opinion about the Intelligence Explosion from all researchers who published at the 2015 NIPS and ICML conferences[…]A total of 352 researchers responded to our survey invitation (21% of the 1634 authors we contacted):
[Assume that HLMI will exist at some point. How likely do you then think it is that the rate of global technological improvement will dramatically increase (e.g. by a factor of ten) as a result of machine intelligence:Median P(…within two years) = 20%Median P(…within thirty years) = 80%]Assume that HLMI will exist at some point. How likely do you think it is that there will be machine intelligence that is vastly better than humans at all professions (i.e. that is vastly more capable or vastly cheaper):Median P(…within two years) = 10%Median P(…within thirty years) = 50%Chance intelligence explosion argument is broadly correctQuite likely (81-100%): 12%Likely (61-80%): 17out even (41-60%): 21%Unlikely (21-40%): 24%Quite unlikely (0-20%): 26%]
Flycer’s explanation for better understanding:
The 2016 Expert Survey on Progress in AI was conducted by AI Impacts, a research organization that studies the long-term impact of artificial intelligence. The survey asked experts in the field to predict when AI would reach certain milestones, such as human-level language translation and the ability to perform a wide range of tasks better than humans. The results showed a wide range of opinions, with some experts predicting these milestones would be reached within the next decade, while others believed it could take several decades or even centuries.
The future of humanity with AGI / HLMI / transformative AI:
[Chance HLMI has positive or negative long run impact on humanity (median answers)Extremely good: 20%On balance good: 25%Neutral:20%On balance bad:10%Extremely bad (e.g human extinction): 5%]
Flycer’s Secondary Explanation:
A survey was conducted to determine the long-term impact of HLMI on humanity. The results showed that 20% of respondents believed it would have an extremely positive impact, while 25% believed it would have a generally positive impact. 10% believed it would have a generally negative impact, and only 5% believed it would have an extremely negative impact, such as human extinction.