My Talk at Foresight 2010: “Don’t Fear the Singularity, but Be Careful: Friendly AI Design” Thursday, Mar 4 2010
friendly ai and videos 11:23 am
Michael Anissimov: “Don’t Fear the Singularity, but Be Careful: Friendly AI Design” at Foresight 2010 Conference from Foresight Institute on Vimeo.
Here’s my talk from Foresight! If you read this blog, there won’t be much new to you. I probably should have summarized the talk at the beginning. Unfortunately I got cut off at around slide 40 out of 55 due to schedule problems, so I missed the opportunity to summarize some of SIAI’s recent work and ended up mainly talking about 1) generic progress in AI, 2) media coverage of AI and Singularity, 3) the intelligence explosion idea, 4) the AI advantage, and 5) the inherent unconnectedness of morality and intelligence (Hume). Ignore the title; I didn’t really get into Friendly AI design at all. It was more of an introduction to why Friendly AI may be required. (I’m not sure I would have even used the term “Friendly AI” if I were making up the talk title again, because it’s been argued by a number of people that the term sounds silly and unserious.)
If I could redo this talk (I plan to do so on video) I would focus a little more on ideas and less on AI advancements, and throw out all the quotes, just quickly summarizing them instead. I would also try even harder to avoid looking down at my laptop during the talk, and would have removed my nametag. I need to buy one of those remote clicker things. I realize I spent a fair amount of time summarizing other people’s AI research rather than ideas unique to me or SIAI, but at the time it seemed necessary because I assumed that few people in the audience would be familiar with the range of advances in AI over the last year alone. People have to understand that AI is making steady progress, otherwise why worry about more advanced AI? If I thought AI really were stuck in the mud, then I wouldn’t be as frantic about the need for safe AI.
Several people pointed out to me that the talk title also seems odd because I am all about getting people to “fear the Singularity” — or fear a negative Singularity where humanity gets steamrolled by indifferent superintelligence. My idea here was that we don’t have to fear the Singularity if we’re careful. I often get the impression that people’s minds just shut down when considering the prospect of an AI Singularity, even if they don’t object to the plausibility of human-level AI in principle, just because they see it as extremely alien in comparison to a human-sparked Singularity. Part of the idea I was going for was that an AI-sparked Singularity can be managed effectively, but as I mentioned, I didn’t even get around to talking about that.
Thinking about my comment on the superficial mundaneity of analyzing the genetic expression of baker’s yeast, I realize that it may not be considered that mundane to some scientists, but I’m not sure because I’m not a biologist that researches microbial genetic expression. I just figured that since yeast is a model organism, we already know a fair amount about its patterns of genetic expression and that the experiments were mainly for show.
You can follow along with the talk with my slides here.




“…so I missed the opportunity to summarize some of SIAI’s recent work…”
Do it in the comments then. Especially if you have something to add in addition to the list of accomplishments I have already seen a few months ago.
Just fyi, for a research project a few years ago I was looking for a cellular environment that was well understood, in terms of chemical elements and reactions. Yeast was one of the best candidates, but there were still lots of unknowns, and proteomics seemed to be making progress on that.
Wow, nice talk.
I try to keep track of a few feeds relating science and future tech on my google desktop sidebar. I don’t have time to investigate all of the latest breakthroughs, and accurately determining what is hype takes time too. This talk was great in that it helped highlight some of the recent important and major developments.
Looking forward to Eve.
Yes, I encourage you to redo the talk on video (and bits of the talk you missed out on), especially if you have prepared for the talk already. Also it would be great for you to video a discussion (perhaps via blogging heads with Eliezer, Aubrey or perhaps even consider *Carl Zimmer) on a) overview of recent AI breakthrough highlights, and b) the summary of SIAIs recent work.
*Carl Zimmer is a biologist, not an AI guy, but he smart, and has written a number of popular books (the one I have read ‘Soul made flesh’ was great). Personally I would like to see him drill down into his views on ideas around Transhumanism and the Singualrity, another flavour to add to the buffet perspectives. See what he had to say on this vid with John Horgan (known Singularity critic) at around 59:00 http://bloggingheads.tv/diavlogs/378
Also I think a talk between Carl and Aubrey would be extremely interesting.
Keep up the good work.
Jay,
I was mainly just going to mention our research team, Visiting Fellows program, Singularity Summit success, Less Wrong, and that’s mainly it. (If you look at the slides I also threw in the Helicopt-o-bot comic at the end.) Our team of researchers, Visiting Fellows and various associates are working on so many projects (4-5 maybe?) that it would take a full 45-60 minutes of talking to summarize it all. I would be interested in doing that, if the researchers are comfortable with sharing their results (many would probably want to wait for journal publication first).
The most recent news about SIAI that I can share now is that a decision theory workshop is happening very soon in the Bay Area. Wei Dai, Nick Tarleton, Eliezer Yudkowsky, and various Visiting Fellows will be participating.
Frank,
Thanks, that’s what I sort of assumed, but I wasn’t sure.
Adam,
Thanks a lot! Yeah, it would be great for Aubrey or Eliezer to talk to Carl Zimmer. His coverage of the Singularity Summit 2009 was the best and most comprehensive I read.
As a public face and voice of the SIAI – and transhumanism in general – you might want to pay a bit attention to your diction. The main issue to work on is not to swallow the t at the end of the word. It’s a bit like reading someone who doesn’t cross their t’s; fully intelligible but it looks a bit funny.
Dunno if you’ve ever paid attention to it, but is that common where you live, or is it just you? Out of all prominent transhumanists, I’ve only noticed this peculiar lack of love for the t’s in your speech.
I think that absolutely friendly AI is impossible as impossible absolutely friendly human. If we design absolutely friendly AI then this AI will be not human-like one. If we want to design human-like AI we need to introduce there motivations, emotions and fuzzy learning and fuzzy changing concepts. So friendliness of such AI will be determined by features of learning by humans and environment like for humans in socity.
So danger of such AI in context of singularity with respect to his more powerful is truly actual problem. And I think that this problem is not avoided or solved.
The mars rovers are entering 3/4th of the way through their 6th year. They do require human intervention.
Air drones have killed THOUSANDS not due to bad intelligence but due to wanton disregard. You should probably avoid saying things that might be inflammatory or make you sound ditzy.