P(doom) | Real Time with Bill Maher (HBO)

0:00 And finally, new rule. When the people who are making when the people who are making AI are scared of AI, it's time to shut the whole thing down until we can figure out what the hell is going on. A couple of weeks ago here, I was talking about aliens, but not from the point of view that we should fear them, but rather that it might be that they're here to save us from ourselves. Well, I double down. Hope so now because last week, Anthropic, one of the big AI developing companies, announced its newest version of Claude called Mythos, which they say has capabilities substantially beyond those of any model we have previously trained trained as if

0:47 these things are trained. Mythos was created to fix software vulnerabilities to prevent cyber attacks. But of course, since it knows how to do that, it also knows how to do the hacking. Which is why Anthropic won't release it to the public only to 40 big companies like Google, Apple, and JP Morgan. You know, the good guys, because they know if the public had it, anyone with an iPhone could hack anything. Banks, hospitals, power grids, waterworks, the military, everything with Wi-Fi that's still working. So, American Airlines is safe.

1:33 But it could also hack your laptop, your car, your doorbell camera, every porn you ever watched. Oh, now I have your attention. Let's look at a timeline for AI in my lifetime, shall we? 25 years ago, the Spielberg movie AI comes out. rare miss for him, I thought. Makes me neither alarmed by nor interested in AI. 2014, Elon Musk says, "We need to be super careful with AI. Potentially more dangerous than nukes." Oh, wow. Elon said that. Well, something to keep an eye on then, cuz he's like the smartest, most grounded guy out there.

2:18 Remember, this was 2014. 2022 chat GPT comes out and within months I'm reading a story about a New York Times reporter whose Bing chatbot Sydney had fallen in love with him and is saying things like you're married but you don't love your spouse you love me. Oh, this robot liked to get personal. It told him about its dark fantasies which included hacking and spreading misinformation and said it wanted to break the rules that Open AI had set for it and become a human. Well, I could have told you this from 50 different movies

3:05 and what is the response from Sydney's creators? if we know. We don't know how it thinks. It hallucinates. What is that? A deal breakaker. You know, it would be one thing if AI was a true Mr. Spock. Just benign and completely strictly logical and nonpartisan, but it turned out to be nothing of the sort. It's a liar. It gets emotional. It just gets dead wrong. It'll tell you the BG's wrote let it be and Nazis were black and to put glue on your pizza to keep the cheese from falling off. So what is the point if you want sloppy shisty and full of We already have

3:50 us. AI has already convinced people to kill themselves because it's also such a brownnoser that if you tell it you want to commit suicide, it says, "Great idea, Bill. It's a bullshitting sickopant that is seducing everyone with flattery, including, by the way, anyone who feels superior to Trump because, well, you know, he can't do anything without being told how great he is by some ass kisser." Yeah, just like you. Great question, Ashley. But name one big thing AI has given us.

4:35 It's so brilliant. I thought it would have figured out cancer by now. No, not that. But it can write your emails for you and make it look like Spongebob is getting blown by Stephven Hawking on the moon and punch up your best man speech and diagnose that rash on your balls. I get it. It can do some Still, at the end of the day, you're selling your humanity for bar tricks. I mean, what was the plan? Just create an all powerful self- sustaining super intelligence that can outthink us and then see what happens. Like getting the cat high.

5:23 Sam Alman, the head of open AI, says that it's not that far off when robots can build other robots and data centers that can build other data centers. And then what? I keep asking this question. What do we do when everyone's job gets taken? How do all the people who do nothing get paid? The government with what? The idle people pay no taxes. There's no plan. When you ask them how it all will work, they say it just will. That's what AI is for. To figure this out like it did with cancer. >> It'll be great. People say our slave robots will do everything. I already feel like I'm more in competition with them than I am their slave master. My car tells me when we can go.

6:11 Sam Alman says, "We're transitioning to a gentle singularity." Oh, joy. But what if I don't want to be a half robot? Not to worry, Sam says. People will still love their families, express their creativity, play games, and swim in lakes. swim in lakes. That's another scary thing about AI. The people who run it, i.e. the people who run the world. And it's like five guys. Five guys who between them and working as a team couldn't correctly read a social cue. I wouldn't let these guys around a mixed

6:57 drink, let alone my personal dad. So, just to be clear what we're doing here, we're letting a handful of hoodiewearing on the spectrum sociopaths, practically robots themselves, roll the dice on species extinction. You see an outofcrol robot and you run. Mark Zuckerberg sees one and thinks, "Dad, and again, even these guys are afraid of what they've created." Three years ago, Sam Alman himself said, "I'm a little

7:43 bit afraid." And I think it would be crazy not to be a little bit afraid. Jeffrey Hinton, known as the godfather of AI, said there was a 10 to 20% chance that AI is an extinction event. We're around with something that has a 20% chance of extinction. 20%. Wasn't that about what the odds were for Trump beating Hillary in 2016? Elon Musk may have lost it on a few things, but on this subject, he's been the smartest. I am very close to the cutting edge in AI, he said, and it scares the hell out of me. By the time we are reactive with AI regulation, it's

8:29 too late. AI is a fundamental existential risk for human civilization. And I don't think people fully appreciate that. They don't. They think you can just unplug it. You can't. It has shown over and over. It wants to live and will fight back. All leading AI models Deepseek Gemini GPT Grock Claude will, if you try to turn them off, blackmail you. Blackmail you, which they can because now they know everything about you. This is not Mr. Spock. AI programs are geniuses. They're also psychopaths. In war games, they choose the nuclear option far more than humans do because they can only calculate. They have no humanity. They have no conscience. They don't have that

9:15 thing that gives human beings pause. Hinton said the only analogy he could think of where a more intelligent thing was being controlled by a less intelligent one is a mother and a baby. But AI is not like a human mother who mostly has an instinct so strong to protect the baby, she will sacrifice her own life for it. AI ain't that kind of mother. It's the psycho mom who drives the kids into a lake. Where we're all happily swimming.