[ExI] Machine Intelligence Survey

[Brent Allsop]

Thank You Anders!!

That was very fascinating information.  It would sure be fun to make 
something like this more dynamic, and increasingly comprehensive, i.e. 
it'd be fun to be able to participate, and see others I know included.

The averages of the 10%, 50%, and 90% ranges seemed to be right on where 
my personal beliefs are.  I thought such a crowd would be much more 
optimistic / aggressive than that so the results were surprisingly 
conservative to me.

Also, the free text, can you think of any milestones, section was 
fascinating.  Open survey questions like this are what canonizer.com was 
designed for.  It would enable the people with many similar ideas to 
colaboratively merge them in a consensus building way and get a 
comparative quantitative measure of just how many people accepted the 
best of such ideas.  Also, it is more open ended so the survey continues 
on indefinitely, allowing anyone interested to contribute, for people to 
jump camps, when their camp is finally falsified, and so on.  So we can 
track things going forward, know, definitively, when an expert consensus 
is achieved, and so on.

Towards that end, I've created a canonizer.com survey topic and started 
a camp of my own, for your milestones question number 4 here:

http://canonizer.com/topic.asp/121

It'd be great to expand such survey's, make them much more 
comprehensive, and track where everyone is as ever more scientific data 
comes in, and be able to use such material as very convincing 
educational references for others on such important still theoretical 
topics.

Anders, would you be interested in cross linking your survey back to 
this, so perhaps any others interested in participating in the future my 
be aware of this possibility at canonizer.com?

Brent Allsop





On 4/22/2011 3:38 AM, Anders Sandberg wrote:
> Here are the results (long delayed) of the survey on AGI we did at the 
> Winter Intelligence workshop:
> http://www.fhi.ox.ac.uk/__data/assets/pdf_file/0015/21516/MI_survey.pdf
>
> Validity questionable, but the results are consistent with earlier 
> surveys - the kind of people who respond to this tend to think 
> mid-century human-level AI is fairly plausible, with a tail towards 
> the far future. Opinions on the overall effect were not divided but 
> bimodal - it will likely be really good or really bad, not something 
> in between.
>