Accuracy of AI Predictions

Chance date bias

There is modest evidence that people consistently forecast events later when asked the probability that the event occurs by a certain year, rather than the year in which a certain probability of the event will

Accuracy of AI Predictions

Updated 4 June 2015 It is unclear how informative we should expect expert predictions about AI timelines to be. Individual predictions are undoubtedly often off by many decades, since they disagree with each other. However their aggregate may still be quite informative. The

Publication biases toward shorter predictions

We expect predictions that human-level AI will come sooner to be recorded publicly more often, for a few reasons. Public statements are probably more optimistic than surveys because of such effects. The difference appears to be less than

Selection bias from optimistic experts

Experts on AI probably systematically underestimate time to human-level AI, due to a selection bias. The same is more strongly true of AGI experts. The scale of such biases appears to be decades. Most public AI predictions

The Maes-Garreau Law

The Maes-Garreau law posits that people tend to predict exciting future technologies toward the end of their lifetimes. It probably does not hold for predictions of human-level AI. Clarification From Wikipedia: The Maes–Garreau law is the statement that “most favorable

AI Timeline predictions in surveys and statements

Surveys seem to produce median estimates of time to human-level AI which are roughly a decade later than those produced from voluntary public statements. Details We compared several surveys to predictions made by similar groups of

Similarity Between Historical and Contemporary AI Predictions

AI predictions from public statements made before and after 2000 form similar distributions. Such predictions from before 1980 appear to be more optimistic, though predictions from a larger early survey are not. Discussion Similarity of predictions over time MIRI dataset