Open Source Artificial Intelligence

Transcript

Chris: My name is Chris LaFleur. I work for congress for representative John Conyers.

A couple of days ago I read about you talking about artificial intelligence and the dangers of it and how as a businessman you are totally against regulation and stuff like that but as a human being you think it is critical that we get ahead of this issue. Can you please elaborate on what are you seeing that we don’t get to see and what as a policy maker I should be looking to do to protect us all.

Elon: It is difficult to appreciate just how far artificial intelligence has advanced and how far it is advancing because we have a double exponential at work. We have an exponential increase in hardware capability and we have an exponential increase in software talent that is going into AI.

Whenever you have a double exponential it is very difficult to predict. Predictions are almost always going to be too conservative in terms of thinking it will be further out than it is.

I don’t know if you’ve seen the videos where you can quite accurately video simulate someone and put words in their mouth that they never spoke. Just Google this, it’s really pretty amazing.

They had so-called generative adversarial network had two of them compete with one another to make the most convincing video so one would generate the video and the other one would identify where it looked fake. The other one would fix that and then go back and forth to the point where you couldn’t tell which was the real video and which was the fake one.

Obviously there have been some very public things like the defeat of Go by AlphaGo the world's best Go champion. People thought defeating Go was never or 20 years away the world’s best Go player was defeated and now that same AlphaGo system can defeat the top best 50 Go players simultaneously with zero percent chance of them winning and that’s one year later. So the degrees of freedom to which artificial intelligence is able to apply itself is really increasing I think by 10 orders of magnitude a year. That’s really crazy. And this is on hardware that is really not well-suited for neural nets. Like a GPU is maybe an order of magnitude better than a CPU, but a chip that is designed optimally for neural nets is an order of magnitude better than a GPU. There are a whole bunch of neural net optimized chips coming out either late this year or next year.

I think part of the role of government is to make sure the public is safe - to take care of public safety issues - I think the right move is to establish some government regulatory agency which at first is just there to gain insight. It’s not about shooting from the hip and putting in rules before anyone knows anything. But you’ve got to set up the agency, it’s got to gain the insight. Once that insight is gained, then start applying rules and regulations.

We have that for aircraft (FAA), we’ve got that for cars, we’ve got that for drugs for food. And I don’t think anyone wants the FAA to go away or the FDA to go away or any of those regulatory agencies. I think we just need to make sure that people do not cut corners on AI safety.

It’s going to be a real big deal and it’s going to come on like a tidal wave.