My interview with Nikola Danaylov for 'Singularity 1 on 1' [link]

by lukeprog1 min read16th Jan 201212 comments

1

Personal Blog

Here. Audio and video available.

12 comments, sorted by Highlighting new comments since Today at 5:26 AM
New Comment

Ugh. Does anyone else think I look smug on camera in this interview? What am I doing with my eyebrows?

Back to mirror-practice for me...

The slight smirk looks a bit smug, but the use of your eyebrows helps in emphasizing what you're saying--It's good nonverbal communication. I think you should be more concerned with your stubble and haircut--not flattering, man.

Yes. Upward-looking camera probably accounts for some of that.

On the upside, your voice sounds great. And the things you're saying don't seem to be too bad so far either :P

The camera angle is very flattering to your chin. That chin could crack mountains !

I think you should have been more critical of the appeal to "Wake, Watch, Wonder" that AI will automatically be good. You could say, for example, that there have been many arguments made that AI will be fundamentally friendly, and every one you know is false, and that there are many good reasons to expect AIs to be indifferent about human life. Or that it sounds like the author was anthropomorphizing AI, which leads you to incorrect conclusions. For example, the claims about evolution and so on only work if the AI is already structured as if it evolved just like humans, but there's no reason at all why that should be the case.

No transcript yet? Reading is so much faster than listening/watching.

Varies. Feynman, a transcript is nice but hearing the guy speak is just brilliant. Some videos, ah, quite the opposite. Luke, which is this more like?

If you own a smartphone, combine a talk or interview with a bike ride or jog. If you moderate the pace right, you can still pay attention while improving personal fitness.

Talks are also useful during car/bus rides where if you are a passenger reading can induce motion sickness.

I don't think he normally produces transcripts. If somebody wants to do that, that'd be great.

I commented on part 3 - as follows:

The case for progressive evolution has been laid out by Robert Wright, Matt Ridley, Peter Corning, Jon Stewart and others. If that includes moral progress - as it certainly seems to - then the future looks bright.

Of course the past is not necessarily a 100% reliable guide to the future. People might be reading the trends wrong. Or we might not yet be out of the woods. Or there might be setbacks. Or the technical problems might be Too Formidable. However, none of those objections looks terribly strong - so there's a pretty good chance that we will make it.

The argument that the target is small is almost completely irrelevant. There are organisations aiming at the target. Optimisers have a long history of hitting small targets. Saying that the target is small means practically nothing.

For the references to Wright, Ridley, Corning, Stewart, see the bottom of this page.

In part 5, Luke gives humanity a 5% chance of surviving this century.

He says of machine intelligence "by default it will be disasterous for humanity".

"By default", a 747 is a pile of scrap. "By default" no aeroplane ever flies. This kind of arguing "by default" just seems to be being used to introduce nonsense into the conversation while making it sound reasonable.

I think this line of argument needs to become uncool. If the best someone has is an argument "by default", then that just indicates that they don't know much about the history of engineering safety, which surely provides a vastly better basis for predicting the success rate of future safety-critical engineering projects, than the whole "by default" line of argument.