Any progress towards AI at all? was Re: [extropy-chat] Futures Past
bpaatsch at bigpond.net.au
Sun Oct 9 22:11:13 UTC 2005
Eliezer S. Yudkowsky wrote:
> Look, I remember when I was dumb enough to think that I could say the
> Singularity had a 90% probability of occurring between 2005 and 2025
> with the median point at 2018. But I haven't been that dumb since I
> read "Judgment Under Uncertainty: Heuristics and Biases", which was, I
> think, around 2003 or so. I just want to note that "people on the SIAI
> lists" could be anyone who hasn't actually gotten themselves kicked off
> as yet. It doesn't mean SIAI endorses the viewpoint.
> I do endorse the viewpoint that we could get transhuman AI within 10
> years, or 5 years, or 1 year, because I don't know all the projects in
> play in detail, and neither do you. I'm rather more skeptical about the
> Friendliness part, I'm afraid.
AI is one area that I haven't had time to even track progress on.
What would be more meaningful to me than projections that AI will
arrive in 10 or 5 or 1 year would be instances given of clear progress,
however small, that actually has been made.
There are faster computers and smaller chips but its not clear to me
that the world is any closer to seeing general AI in 2005 than it was in
The impression that I get is that pretty much all that has been built up
by AI researchers is a list of ways that don't work and approaches that
have not yeilded any gain.
Can you think of any clear instances of positive *progress* towards the
goal of building an AI?
More information about the extropy-chat