LESSWRONG
LW

Personal Blog

22

Self-assessment in expert AI predictions

by Stuart_Armstrong
26th Feb 2013
1 min read
9

22

Personal Blog

22

Self-assessment in expert AI predictions
13Kaj_Sotala
0KatjaGrace
0Kaj_Sotala
4gwern
4Vladimir_Nesov
2Stuart_Armstrong
2Mitchell_Porter
0TrE
0Stuart_Armstrong
New Comment
9 comments, sorted by
top scoring
Click to highlight new comments since: Today at 9:50 PM
[-]Kaj_Sotala12y130

Thanks for making the post, Stuart.

In case anyone's wondering, this post doesn't mention the surveys from AGI conferences because anyone participating in those will be extremely self-selected. XiXiDu's interview series is somewhat better, in that he also contacted names in mainstream AI, but he also spoke to a number of folks who were specifically chosen for working in the field of AGI so that's not a representative sample either.

Reply
[-]KatjaGrace11y00

Do you mean the people at AGI conferences are extremely self-selected in the sense of them choosing to be in AGI or to attend the conference, or do you mean that within the conference, there is strong selection regarding who chooses to do the surveys?

Reply
[-]Kaj_Sotala11y00

The former.

Reply
[-]gwern12y40

Seems in line with Tetlock about experts and media appearances, no?

Reply
[-]Vladimir_Nesov12y40

The images don't load for me. (In general, it's better to use LW's inbuilt hosting, so that the images are maintained in the same archive and don't disappear later.)

Edit: Fixed now, thanks!

Reply
[-]Stuart_Armstrong12y20

Edited that. Is it better now?

Reply
[-]Mitchell_Porter12y20

From the title, I thought this might be about an AI's "self-assessment" and the role this would play in predictions involving itself (e.g. perhaps predictions of how it would think and act after a particular modification).

Reply
[-]TrE12y00

More than 50 years: 26 years

*experts

[This comment is no longer endorsed by its author]Reply
[-]Stuart_Armstrong12y00

Thanks!

Reply
Moderation Log
Curated and popular this week
9Comments

This brief post is written on behalf of Kaj Sotala, due to deadline issues.

The results of our prior analysis suggested that there was little difference between experts and non-experts in terms of predictive accuracy. There were suggestions, though, that predictions published by self-selected experts would be different from those elicited from less selected groups, e.g. surveys at conferences.

We have no real data to confirm this, but a single datapoint suggests the idea might be worth taking seriously. Michie conducted an opinion poll of experts working in or around AI in 1973. The various experts predicted adult-level human AI in:

  • 5 years: 0 experts
  • 10 years: 1 expert
  • 20 years: 16 experts
  • 50 years: 20 experts
  • More than 50 years: 26 experts

On a quick visual inspection, these results look quite different from the distribution in the rest of the database giving a much more pessimistic prediction than the more self-selected experts:


But that could be an artifact from the way that the graph on page 12 breaks the predictions down to 5 year intervals while Michie breaks them down into intervals of 10, 20, 50, and 50+ years. Yet there seems to remain a clear difference once we group the predictions in a similar way [1]:

This provides some support for the argument that "the mainstream of expert opinion is reliably more pessimistic than the self-selected predictions that we keep hearing about".

[1] Assigning each prediction to the closest category, so predictions of <7½ get assigned to 5, 7½<=X<15 get assigned to 10, 15<=X<35 get assigned to 20, 35<=X<50 get assigned to 50, and 50< get assigned to over fifty.

 

Mentioned in
48Original Research on Less Wrong