Tesla and SpaceX CEO Elon Musk, who has exposure to the most cutting-edge AI, spoke to more than thirty of the nation’s governors on Saturday. You could hear a pin drop when he issued a dire warning about AI.
“It was quite astounding to see him speak, saying, ‘We really need to do something about AI.’ He says that AI is going to change the world, and not in a positive way,” Glenn said Monday on radio.
Robots will ultimately be able to do everything better than humans, literally introducing an entirely new way of thinking and solving problems.
“I’m not exactly sure what to do about when machines reach their potential. This is really like the scariest problem to me. I lose a lot of sleep,” Glenn said.
Various technology companies are racing to build better AI systems to stay competitive in the market, without necessarily examining the long-term impact.
“How are we going to teach AI absolute truth, when we don’t even agree with absolute truth? We have no boundaries on anything. If you can do it, you should do it. There is no such thing as good pain. There is no such thing as good failure in today’s society. So what do we become?” Glenn asked.
Musk additionally warned that the development of AI must be regulated, while development proceeds slowly and cautiously.
FABRICE COFFRINI/AFP/Getty Images
GLENN: Did you guys see what Elon Musk said at the — the mayor’s conference this weekend? Or the governor’s conference. It was quite astounding to see him speak, saying, “We really need to do something about AI.” He says that AI is — is going to change the world, and not in a positive way. He says that AI is — is a fundamental — quoting, a fundamental risk to civilization. He says it’s a fundamental risk to the existence of mankind.
On the artificial intelligence front, I have exposure to the very most cutting edge AI. And I think people should really be concerned about it. I keep sounding the alarm bell, but until people see like Roberts going down the streets killing people, they won’t know how to react because it seems so ethereal. He said we should really be concerned about AI and that emerging technology presents a rare case where the government should be proactive in regulation instead of reactive to industry. By the time we’re reactive to AI, regulation will be too late.
Normally, the way regulation is set up, a whole bunch of bad things happen. There’s a public outcry. And then after years, a regulatory agency is set up to regulate that industry. There’s a bunch of opposition from companies who don’t like to be told what to do by regulators, and it takes forever. That, in the past, has been bad, but not something which represented a fundamental risk to the existence of civilization.
Car accidents, airplane crashes, faulty drugs, bad foods are merely harmful to certain parts of society. But AI will affect all of society. I’m against overregulation for sure, but I think we better get — we better get that with AI. I really think we need a lot of government regulation here.
Robots will be able to do everything better than us. I’m not — I’m not exactly sure what to do about when machines reach their potential. This is really like the scariest problem to me. I lose a lot of sleep.
PAT: Clearly. He talks about it all the time.
GLENN: All the time. It’s really why he’s going to Mars.
PAT: Yeah, yeah.
GLENN: I mean, most people don’t really realize that. That’s why he says —
PAT: Escaping AI essentially.
GLENN: Yeah, he says this threat is so bad that if humans aren’t on another planet, they risk extermination from AI.
STU: Every Elon Musk speech is like the — as the opening music is running in some apocalyptic story. They’re showing these old-timey clips of some guy predicting terrible things, and then you just join the present day with like 12 people left on the planet. It really is like that.
I mean, he’s freaked out beyond even you, I think, Glenn. Isn’t he? Isn’t he? I mean, you don’t even — I know you’ve got fears and concerns about AI.
GLENN: I am really concerned about AI.
PAT: And I have some too. But I’m not like Elon Musk over it.
GLENN: The problem with AI is it doesn’t think like us. And that’s what we cannot predict. And we just learned this from — what is that? Myjong or something? What is that game, the Chinese game that’s harder than chess?
PAT: Yeah, it’s more complex than — I don’t remember. But it beats the masters at that game.
GLENN: Okay. So AI just — remember, when Big Blue beat, what was it, Kasparov? He accused IBM of cheating. And they’re like, “No, it’s AI.” But it took them like three years to beat Kasparov. So there’s another game. This Chinese game — you’re looking it up, Pat?
GLENN: And it just beat the master. The Grandmaster of this — of this game. It’s supposedly far — far more complex than chess.
It beat the master, and the master said, “It’s an illegal move. You can’t make that move.” And they started going back and looking at it. And it’s not an illegal move. It’s just a move that no human had ever thought of. And it won the game quickly.
It apparently — they’re saying now that AI can beat the top 50 masters all combined at this game.
PAT: Is it Kiji?
GLENN: Maybe. I don’t know. I haven’t played it in a while.
STU: What do you think about his approach there? It’s interesting in that, you know, Elon Musk is a business guy, but certainly very friendly with government. A big — gets chunks of money every time he sells one of his products from the government. Also, is a fan — big green energy guy. Certainly is on board for overregulation, even despite that last claim. At least in my opinion.
STU: However, if you go and you overregulate artificial intelligence, you wind up with, what? Japan, China, Russia? Who is leading the way instead of us? We’re not going to be able to regulate them.
GLENN: Here’s the problem: It needs to be a global —
STU: Global governance though is what that —
GLENN: Hang on just a second. We have to come up with something like in the movie I, Robot. You know, the three principles. We have to come up with that.
Go back to the game. It doesn’t think like we do. So they’re saying now that AI almost has to be introduced to humans like a baby. And it’s because you will say, wait. Watch the furniture. And you will say the same thing. Hey, hey, be careful of the baby.
And it won’t know the difference between the baby and the furniture. It doesn’t know — it doesn’t know. So you have to teach it the value of everything that it’s going to encounter.
Well, we don’t even know the value of life. We’re having that discussion now. Life is sacred if…
STU: Unless you’re too young or too old or not wanted.
GLENN: Too sick.
STU: Or too sick. Too poor?
GLENN: Or don’t have a life worth living.
STU: Yeah. How many times have we heard, you have to have X, Y, or Z government program or life isn’t worth living.
GLENN: Right. So they’re saying that it could quickly learn in a completely different way than we’re expecting.
PAT: Yeah. And that’s what happened — the game is called Go.
GLENN: That’s what it’s called.
PAT: So this machine called Alpha Go was taught how to play — and at first it lost. And then — but it learned. And it developed this unusual style of playing that nobody else played like.
PAT: So that’s how it started. And the Grandmaster who was only 19 years old tried to play in the same unusual style and still lost. So a little interesting.
GLENN: Yeah, you want to talk about — you want to talk about asymmetrical warfare, it’s AI.
GLENN: AI will have — will be so asymmetrical to our puny little brains and the way we think. And it will be — it will literally be introducing alien thinking to the world. It will not think like us. It will learn and think in an entirely new way, which will be great —
STU: There will be a lot of positive ramifications of that too, by the way.
GLENN: It will be great.
STU: Things that have stumped us like cancer cures. And perpetual motion machines.
GLENN: Yes. Correct. But think we can’t even as a society agree on basic things like what is truth.
How are we going to tell — teach AI absolute truth, when we don’t even agree with absolute truth. We don’t even know what absolute truth — we have no boundaries on anything. If you can do it, you should do it.
There is no such thing as good pain. There is no such thing as good failure in today’s society. So what do we become?
Elon Musk said, the transportation job sector accounts for 4.6 million jobs. It’s going to be the first thing to go fully autonomous. The robots will do everything, bar nothing.
Various technology companies are racing to build better AI systems to stay competitive in the market. Facebook, with Messenger, Apple with Siri, Amazon with Alexa. Some major consumer examples around the corner. AI-powered autonomous cars. Service companies like Uber, that is increasingly waving AI into its ride-hailing app. Musk said we really need the regulators to come in and say, “Hey, guys, we need you to take a pause and make sure we’ve thought this thing through. When it’s cool and the regulators are convinced it’s safe to proceed, then you can go. But otherwise, slow down.”
Complicating matters is the fact that the biggest companies developing AI right now are publicly held. And stock owners will demand the company aggressively stay ahead of the competition when it comes to developing AI.
STU: I don’t know how you — I don’t know how you can regulate this.
GLENN: You can’t.
JEFFY: You can’t.
STU: You can try to do something where you come up with best practices among the industries involved and try to set down those basic guidelines. I mean, you could try to treat them like nuclear weapons. That’s the only other type of thing — and, again, even that is obviously not easily self-controlled. But this stuff will start creating itself. Once it’s successful, it’s going to advance on its own, and you’re not going to be able to stop it.
GLENN: Here’s what he said: You’re going to see robots that will learn to walk from nothing, within hours.
PAT: That would be something. That —
GLENN: He said, while killer robots are easy to visualize, the most dangerous threat is the deep intelligence in the network. He said they could start a war by creating fake news and spoofing email accounts and sending out fake press releases as a way to manipulate information. The pen today is mightier than the sword.
GLENN: How are you going to outthink Deep Mind. How are you going to — how are you going to prove that something — remember, the Russians just perfected the picture of Neil Armstrong on the moon, except the American flag was replaced with a Russian flag. And the American flag was — on his shoulder, was replaced with CCCP. And they offered I think $100,000 to anyone who could prove that that had been manipulated.
They now say that they have perfected artificial manipulation of pictures, where you can’t tell that that is not a real photo. That no investigation can tell that that’s been digitized and re-created. Changed. How are you — I mean, do you remember back in the ’90s: You won’t believe your eyes. You won’t believe in anything anymore. Because everything will be fake and manipulated. And you won’t know what’s real or not.
STU: What I’m starting to fear is the actual willing start of that. In that —
STU: There’s so much crap out there anymore, I find more and more difficult to care about it. Like, my answer to everyone lying all the time is to just not believe anybody and just — just blow it off. Which as a society is not a good reaction. Hopefully I’m the only doing that.
STU: But I don’t think I am.
JEFFY: No, you are not. No, you are not.
STU: I do not think I am. You just wind up realizing more and more, so little of this stuff winds up mattering in your life. That it’s easy to just kind of blow it off. Then there’s nobody guarding the gate.