Latest / Elliot In The Morning / EITM: Existential Threat 9/10/26
Transcript
- 0:00Speaker 1: Did everybody see the story about Jacob Coxon? Jacob Coxon
- 0:05Speaker 1: used to work for, he just resigned from his job
- 0:09Speaker 1: at Anthropic.
- 0:11Speaker 2: Oh, this guy.
- 0:11Speaker 3: Yeah, right.
- 0:12Speaker 1: So he worked at Anthropic. He previously had worked at
- 0:15Speaker 1: open AI. And his big thing yesterday was that, what
- 0:23Speaker 1: is it within the next decade? AI will kill all humans.
- 0:30Speaker 1: Am I naive that I just dismissed that completely?
- 0:33Speaker 4: You think it's completely not possible?
- 0:34Speaker 1: 100%.
- 0:34Speaker 3: So I kept seeing that. Yes.
- 0:38Speaker 2: But did he go on? Did he mean kill like
- 0:40Speaker 2: off livelihood of humans or actual death and destruction?
- 0:45Speaker 3: He said, okay, here's what he said.
- 0:49Speaker 1: People building AI earnestly believe it could kill us all
- 0:54Speaker 1: by the end of the decade. I'm sorry, not in
- 0:57Speaker 1: a decade, by the end of a decade. I took
- 1:02Speaker 1: that as kill us.
- 1:05Speaker 2: Yeah, that's how I'm now interpreting it.
- 1:07Speaker 1: He said, do not underestimate the power of this technology.
- 1:12Speaker 1: These will soon be superhuman systems that can hack anything,
- 1:17Speaker 1: revolutionize any field overnight and.
- 1:20Speaker 3: Acquire real power and resources.
- 1:24Speaker 1: So then they went around like a bunch of, a
- 1:26Speaker 1: bunch of like people who cover technology and, went around
- 1:31Speaker 1: and started asking other experts.
- 1:36Speaker 3: Is he on anything?
- 1:38Speaker 2: And a lot of.
- 1:38Speaker 3: Experts were like, yeah, you know what? I give it
- 1:40Speaker 3: about a 10, 20% chance of him being right.
- 1:43Speaker 2: Oh, him being right. They were also saying it's between
- 1:47Speaker 2: 10 and maybe 20%.
- 1:48Speaker 1: No, no, like of him being right, that he could
- 1:51Speaker 1: kill all humans by the end of a decade, by
- 1:54Speaker 1: the end of this decade.
- 1:57Speaker 2: But you fear not.
- 1:59Speaker 3: I dismiss it right out of hand.
- 2:02Speaker 2: And his resignation doesn't– or his quitting or however he
- 2:06Speaker 2: left the company, that doesn't add to the concern at
- 2:10Speaker 2: all for you?
- 2:11Speaker 3: No.
- 2:12Speaker 1: No. I mean do I believe– do I believe that
- 2:17Speaker 1: he believes it?
- 2:18Speaker 3: Yes, I do.
- 2:19Speaker 1: Like I believe that he believes it, that it could happen.
- 2:24Speaker 1: And not– and I'm not saying like this guy is crazy.
- 2:28Speaker 1: I just don't believe it happens. I don't believe it
- 2:30Speaker 1: can happen. I believe it would be stopped before it
- 2:33Speaker 1: can happen. We're not going to let that happen. Do
- 2:36Speaker 1: I believe that there are evil, evil, evil people working
- 2:41Speaker 1: in the field and working on AI?
- 2:43Speaker 2: Of course.
- 2:44Speaker 3: What have I always said?
- 2:45Speaker 1: Show me a great invention and I'll show you somebody
- 2:48Speaker 1: who's using it for evil.
- 2:50Speaker 3: So do I believe that exists?
- 2:52Speaker 5: Of course.
- 2:54Speaker 1: But there's too much good to allow that to happen.
- 2:57Speaker 1: But they're also saying that you can't control it.
- 3:00Speaker 3: Exactly.
- 3:01Speaker 1: Right.
- 3:01Speaker 3: Which I also don't believe.
- 3:03Speaker 2: You hear often how these agents have gone rogue within
- 3:07Speaker 2: these companies.
- 3:08Speaker 3: But what does that mean?
- 3:09Speaker 1: Like, all of a sudden it started, like, hacking stuff
- 3:13Speaker 1: you didn't want it hacked.
- 3:14Speaker 3: And what do they do? You know what? Like, don't
- 3:15Speaker 3: you feel like you could go, oh, watch this. Unplug.
- 3:17Speaker 2: Done.
- 3:19Speaker 3: Seriously.
- 3:20Speaker 4: It doesn't. It seems, like, weird to say it would
- 3:23Speaker 4: be that easy to stop it in its tracks.
- 3:27Speaker 3: Blow it up.
- 3:30Speaker 2: So if we talk to, because he was a researcher.
- 3:35Speaker 3: Correct.
- 3:36Speaker 2: He was at Anthropic or was he OpenAI?
- 3:38Speaker 5: Both.
- 3:39Speaker 3: He started at OpenAI, went to Anthropic.
- 3:41Speaker 1: His most recent job was Anthropic.
- 3:44Speaker 2: What would someone who's working quite a bit with AI
- 3:50Speaker 2: say in response to these claims? So not someone who's
- 3:54Speaker 2: developing the intelligence, but there's But anybody who's.
- 3:58Speaker 3: Using chat GPT isn't going like, well, they're going to
- 4:01Speaker 3: take it and kill all humans.
- 4:03Speaker 1: I use chat GPT and I don't even think it's
- 4:05Speaker 1: going to kill me.
- 4:06Speaker 2: I'm not talking about using it as an assistant occasionally
- 4:11Speaker 2: every other day. I'm saying that works with it pretty
- 4:14Speaker 2: heavily or is encouraged by their businesses to work with it.
- 4:18Speaker 3: What would they say? Well, some of them are saying
- 4:22Speaker 3: that it's right.
- 4:23Speaker 2: That sounds like they're on the same level as this
- 4:26Speaker 2: Jacob guy.
- 4:26Speaker 1: Correct.
- 4:27Speaker 3: Correct.
- 4:28Speaker 2: I'm talking about somebody who's being told by bosses. I'm
- 4:33Speaker 2: not saying it has to be an AI company.
- 4:35Speaker 3: Right.
- 4:36Speaker 2: Who just spends a lot of their day interacting with
- 4:39Speaker 2: artificial intelligence.
- 4:41Speaker 3: How's it going to kill me?
- 4:43Speaker 2: That's what I'm wondering.
- 4:45Speaker 3: Like, I don't have a humanoid.
- 4:48Speaker 1: You think I'm going to have a humanoid by 2030
- 4:49Speaker 1: that's going to pick up a knife and stab me
- 4:51Speaker 1: to death?
- 4:52Speaker 3: Zero chance.
- 4:53Speaker 2: Like, do they have these same concerns? And every single
- 4:56Speaker 2: day when they're working with it, do they feel like, wow,
- 4:58Speaker 2: we are really training this to a degree that might
- 5:01Speaker 2: not be controllable?
- 5:03Speaker 1: Or am I thinking of it the wrong way? Like,
- 5:05Speaker 1: I have no fear a humanoid's going to stab me.
- 5:07Speaker 3: No, I really don't. Or shoot me.
- 5:09Speaker 1: Or choke me. Or garrote me. I'm not afraid of
- 5:13Speaker 1: a humanoid. I mean, if I'm confronted with one on
- 5:16Speaker 1: the street, it'd probably kick my ass, but I'm not...
- 5:19Speaker 1: Or am I thinking of it wrong? Where they're not
- 5:22Speaker 1: going to pick up a knife and stab you, but
- 5:24Speaker 1: what they are going to do is poison the entire
- 5:27Speaker 1: water system.
- 5:28Speaker 2: Yeah, exactly. Yeah, that's a great example.
- 5:32Speaker 1: Then, yeah, they probably can't. But would it be done intentionally? No,
- 5:39Speaker 1: they're saying yes, that it would be done intentionally.
- 5:41Speaker 2: So you're saying... not so much someone using AI for evil,
- 5:47Speaker 2: AI just becoming evil.
- 5:50Speaker 1: Yes, he says, the Coxon guy, neither company, talking about
- 5:56Speaker 1: anthropic and open AI, neither company is acting responsibly. They
- 6:01Speaker 1: are racing straight to self-improving super intelligence.
- 6:09Speaker 2: But isn't this a little bit, like what Bill Gates
- 6:12Speaker 2: said years ago, and he's sort of changed his tune
- 6:15Speaker 2: on it. Now, of course, AI is probably enriching him further,
- 6:20Speaker 2: but you heard worry of this before Jacob Coxon came out.
- 6:27Speaker 2: Did someone poison your food?
- 6:29Speaker 6: No, but.
- 6:29Speaker 4: No, that crouton hitting the plate was a little loud.
- 6:31Speaker 3: No, it's weird, right?
- 6:32Speaker 4: That's from the Caesar pizza.
- 6:34Speaker 3: I've never had a crouton on a pizza. I'm eating
- 6:36Speaker 3: hard bread on crushed.
- 6:38Speaker 2: So you cracked a tooth.
- 6:39Speaker 3: No, that was just one of the.
- 6:40Speaker 4: That was the crouton hitting the plate.
- 6:42Speaker 2: Oh, it did fall out of your mouth.
- 6:43Speaker 3: No, it fell off the pizza going through my mouth.
- 6:46Speaker 5: Oh, oh, oh.
- 6:48Speaker 3: I've never had bread on bread. That's fantastic. I mean,
- 6:51Speaker 3: let's count a sandwich.
- 6:52Speaker 6: Yeah, you.
- 6:56Speaker 3: But yes, Bill Gates did say that. And then did
- 6:59Speaker 3: change his tune. You're absolutely right.
- 7:01Speaker 2: But maybe he wasn't saying such dire things and just
- 7:04Speaker 2: speaking more about artificial intelligence impact on the workforce. I
- 7:08Speaker 2: don't remember exactly what he said.
- 7:10Speaker 3: But it's like one thing to go. Can AI hack
- 7:13Speaker 3: my computer? Yeah, probably.
- 7:15Speaker 5: Yeah.
- 7:15Speaker 3: A hundred percent. But is it going to kill me?
- 7:22Speaker 1: But also what would be like, like the people that are,
- 7:26Speaker 1: and again, I do understand that there are evil people,
- 7:29Speaker 1: but what does that get you?
- 7:33Speaker 2: People do it for the thrill.
- 7:34Speaker 3: Of killing everybody?
- 7:37Speaker 2: But what if you're totally against an entire nation? It
- 7:43Speaker 2: may not be someone seeking for global destruction, but targeted, sure.
- 7:48Speaker 3: Yeah, no, I guess so.
- 7:50Speaker 2: But again, I interpret it more, and it's mostly from
- 7:52Speaker 2: what you're saying, because all I saw was the big headline.
- 7:55Speaker 3: Right.
- 7:56Speaker 2: I do think it's less the use of AI and
- 8:00Speaker 2: the utilization of AI. It is just The artificial intelligence acting.
- 8:06Speaker 3: Yeah, either way, we're all going to be dead in
- 8:08Speaker 3: three years.
- 8:12Speaker 2: Well, no, it wasn't three years.
- 8:14Speaker 3: By the end of the decade.
- 8:16Speaker 1: Oh.
- 8:17Speaker 3: Yeah, by the end of the decade.
- 8:20Speaker 2: Never mind.
- 8:21Speaker 1: People at Anthropic and OpenAI are gambling with our lives.
- 8:27Speaker 1: I earnestly believe that it could kill all of us
- 8:30Speaker 1: by the end of the decade.
- 8:33Speaker 2: Just got so much more real.
- 8:35Speaker 5: Right.
- 8:35Speaker 1: That's why I'm eating. I don't have a lot of
- 8:38Speaker 1: years left to eat. Where am I going, Kristen?
- 8:42Speaker 5: Line six?
- 8:43Speaker 4: Oh, you're going to ramp it up these next three years.
- 8:45Speaker 1: Or am I just being naive where I go, this
- 8:48Speaker 1: is so impossible. Like, I appreciate the Henny Penny, the
- 8:53Speaker 1: sky is falling. And I would even go as far
- 8:56Speaker 1: as to say, I believe, be careful. Like, I'm sure
- 9:00Speaker 1: they've been able to do a ton of good with it.
- 9:05Speaker 1: But you can see some effects where it's been very disruptful.
- 9:10Speaker 2: Disruptive?
- 9:12Speaker 3: No, disruptful.
- 9:14Speaker 2: So is it you being naive or is it more ignorance?
- 9:20Speaker 3: If I could choose a word, I'd rather say I'm
- 9:21Speaker 3: naive than ignorant. Somebody already called me stupid earlier. Hi,
- 9:27Speaker 3: Elliot in the morning.
- 9:29Speaker 6: Hey, this is me.
- 9:30Speaker 3: Yeah, hi, who's this?
- 9:33Speaker 5: Hey, this is Corey. So, in 2025, I was in
- 9:36Speaker 5: the top 1% of open AI users. So, I might
- 9:39Speaker 5: have an opinion for you.
- 9:41Speaker 3: Okay, go ahead.
- 9:42Speaker 5: So, I think he could be very possibly right. It
- 9:50Speaker 5: certainly has the capability to it. From a lot of
- 9:53Speaker 5: the research I've done recently, I I agree with my
- 9:55Speaker 5: main concern is that anthropic and open AI. And I
- 9:59Speaker 5: wouldn't be surprised if, uh, whoever runs Grog, they're all
- 10:04Speaker 5: racing to super intelligence. It's a big race towards it
- 10:08Speaker 5: because whoever gets super intelligence first becomes the first multi-trillionaire
- 10:12Speaker 5: and surpasses everyone.
- 10:14Speaker 1: Right. So, I mean, obviously, I mean, I get that
- 10:16Speaker 1: greed would be a huge driver that I totally get.
- 10:20Speaker 5: But I think safety concerns are being just blatantly overlooked
- 10:24Speaker 5: or ignored. I mean, the models they have available internally
- 10:27Speaker 5: that haven't been released yet. I believe that they're afraid of,
- 10:31Speaker 5: which is where you get people stepping down and being
- 10:35Speaker 5: pushed aside for not wanting to continue that work.
- 10:38Speaker 3: How is it going to kill all of us?
- 10:41Speaker 1: Is it going to be? Is it going to be?
- 10:45Speaker 1: Is it going to be done? intentionally by somebody who's
- 10:49Speaker 1: using it for evil? Or is it going to be
- 10:51Speaker 1: like in one of these where it goes rogue and
- 10:54Speaker 1: there's just no stopping it?
- 10:57Speaker 5: No, I think it's going to go rogue and there's
- 10:58Speaker 5: no stopping it. I mean, not to sound exactly like
- 11:01Speaker 5: a movie, but I mean, if you look at it,
- 11:02Speaker 5: humans are the problem with the planet.
- 11:05Speaker 3: No, we're not actually.
- 11:08Speaker 4: Okay. Meaning it's us who's destroying it.
- 11:10Speaker 3: I mean, we're creating something that's going to kill all
- 11:13Speaker 3: of us.
- 11:14Speaker 1: But no, we're far better with people here than not here.
- 11:18Speaker 1: Are we?
- 11:19Speaker 5: Yes. I don't think he agrees.
- 11:23Speaker 2: All right, tomato, tomato.
- 11:25Speaker 7: Yeah, no, I mean, I.
- 11:26Speaker 2: No, that sounds pretty black or white.
- 11:28Speaker 1: I think we are. Yeah, no, I'd much rather. I'd
- 11:31Speaker 1: much rather have people here.
- 11:33Speaker 5: I mean, me too, but that doesn't necessarily mean that
- 11:35Speaker 5: we're good for... our end goal, which, I mean, none
- 11:39Speaker 5: of us can even definitively say what it is. We've
- 11:41Speaker 5: been arguing about that for how long?
- 11:43Speaker 3: What, for our end goal?
- 11:44Speaker 1: I just want to be happy and live a good life.
- 11:46Speaker 3: Pizza. The amen, right?
- 11:49Speaker 1: Yeah, no, I understand what you're saying, but you think
- 11:53Speaker 1: one of these things goes rogue, and then that's it?
- 11:58Speaker 5: I mean, not that that's it, but, I mean, you
- 12:01Speaker 5: got to think about it. In 2026, we had our
- 12:03Speaker 5: first confirmed AI kill. Now, it was military purposes, but
- 12:08Speaker 5: the point being is look at what Anthropic and OpenAI
- 12:12Speaker 5: are doing. What no one's talking about is what is
- 12:14Speaker 5: DARPA doing right now?
- 12:16Speaker 3: What is who doing?
- 12:18Speaker 7: DARPA.
- 12:19Speaker 5: DARPA. We're 10 to 15 years ahead of everybody. Right. So,
- 12:23Speaker 5: I mean, yeah, we got AI, what, five, six years ago,
- 12:28Speaker 5: kind of a base level, and it's growing. The military
- 12:32Speaker 5: and U.S. government and probably other governments throughout the world
- 12:34Speaker 5: have had access to this and had this technology.
- 12:36Speaker 3: For 20, 30 years.
- 12:38Speaker 5: There's no telling where they're at.
- 12:39Speaker 1: So am I just too naive to even fathom that
- 12:46Speaker 1: it's capable of going rogue and killing everybody in the
- 12:50Speaker 1: next handful of years?
- 12:51Speaker 5: I mean, I don't know if I would call it naive.
- 12:55Speaker 5: I mean, I think there's certainly a lot of material
- 12:57Speaker 5: out there. The farther you get into it and, you know,
- 13:00Speaker 5: depending on what rabbit holes you want to go down,
- 13:01Speaker 5: you can certainly come to that conclusion.
- 13:03Speaker 1: Right.
- 13:04Speaker 5: And I don't think it's one of those things where
- 13:05Speaker 5: we're building a machine that just wants to take over
- 13:07Speaker 5: the world and kill everyone. But I think it's certainly
- 13:11Speaker 5: human nature and greed that we're going to do that
- 13:14Speaker 5: on accident and give it access and resources to where
- 13:17Speaker 5: we can't stop it.
- 13:20Speaker 3: That's kind of what there was another.
- 13:22Speaker 5: We'll be safe.
- 13:23Speaker 3: There was another. A thank you, sir. Thank you.
- 13:26Speaker 1: There was another expert, the one that came out and
- 13:28Speaker 1: said like, oh yeah, 10, 20%, this guy's right.
- 13:32Speaker 3: He came out and said, Jacob is correct here.
- 13:34Speaker 1: We really do earnestly believe AI could kill all humans.
- 13:39Speaker 1: I personally think it's just greater than 10% within the
- 13:42Speaker 1: next decade. I believe Anthropic is trying its best, but
- 13:47Speaker 1: we do not yet have a plan to solve alignment
- 13:50Speaker 1: for super intelligence and are not clearly on track to.
- 13:56Speaker 1: In the AI industry, alignment refers to the work by
- 14:00Speaker 1: AI developers to ensure that the system behaves in accordance
- 14:04Speaker 1: with human values and intentions.
- 14:08Speaker 5: Right.
- 14:10Speaker 2: Why? So I keep being messaged paperclip emojis.
- 14:14Speaker 4: What?
- 14:15Speaker 3: Oh, from... Go ahead. Enough people think I'm dumb today.
- 14:21Speaker 5: Go ahead.
- 14:23Speaker 2: Well, I may have to... it even closer to you.
- 14:27Speaker 2: I don't get this.
- 14:28Speaker 3: The paperclip emoji?
- 14:29Speaker 2: Is that a reference to something with AI? Or is
- 14:34Speaker 2: that like a known AI symbol of artificial intelligence?
- 14:38Speaker 1: My guess was going to be that it is a
- 14:42Speaker 1: flashback to if, like, early on, the paperclip was what
- 14:48Speaker 1: you asked Microsoft for help.
- 14:52Speaker 3: That was, I mean, in its earliest form.
- 14:57Speaker 8: 90s AI.
- 14:57Speaker 3: Yeah, no, it was. And look where it's come since then.
- 15:01Speaker 1: I could be wrong, but I would feel like, and
- 15:03Speaker 1: I don't live in Silicon Valley. However, I bet that's
- 15:08Speaker 1: a throwback to, oh yeah, and people who saw this
- 15:11Speaker 1: never thought.
- 15:13Speaker 3: The paperclip would turn into what it is now. Now,
- 15:16Speaker 3: go ahead and tell me I'm wrong.
- 15:17Speaker 2: The paperclip problem is a thought experiment. that shows how
- 15:22Speaker 2: a super intelligent AI with a simple harmless goal could
- 15:26Speaker 2: accidentally destroy humanity. It is told, so the machine is
- 15:34Speaker 2: told to maximize paperclip production, but in time, decides it
- 15:40Speaker 2: needs to turn all matter on Earth into paperclips, including humans.
- 15:45Speaker 3: Okay, how are they going to do that? How are
- 15:46Speaker 3: you going to turn it? That's where I go, all right,
- 15:49Speaker 3: this is dumb. How are they going to turn a
- 15:53Speaker 3: human into a paperclip?
- 15:55Speaker 1: Can't.
- 15:56Speaker 4: Until it spews out the answer.
- 16:01Speaker 1: Didn't AI just solve the most impossible math problem of
- 16:05Speaker 1: all time?
- 16:05Speaker 3: Yeah, I saw that. Yeah.
- 16:09Speaker 2: So suppose, here's more on that. This paperclip thing, I've
- 16:13Speaker 2: never heard of me talking about it. It's not jogging
- 16:16Speaker 2: your memory?
- 16:17Speaker 3: No, no, clearly I'm way off.
- 16:19Speaker 2: Suppose we have the AI and the only goal is
- 16:22Speaker 2: to make as many paperclips as possible. The AI will
- 16:25Speaker 2: quickly realize that it would be much better if there
- 16:27Speaker 2: are no humans because humans might decide to switch it off.
- 16:32Speaker 3: Right. That's me.
- 16:33Speaker 2: I want to unplug it. You said even blow it up.
- 16:37Speaker 2: Because if humans do so, there would be fewer paperclips. Also,
- 16:41Speaker 2: human bodies contain a lot of atoms that can be
- 16:43Speaker 2: made into paperclips. The future that the AI would be
- 16:47Speaker 2: trying to gear towards would be one in which there
- 16:49Speaker 2: would be a lot of paperclips, but no humans.
- 16:54Speaker 3: Great theory. You know what's not going to become a paperclip?
- 16:58Speaker 3: This guy. Where am I going, Kristen? Line two? Hi,
- 17:06Speaker 3: Elliot in the morning.
- 17:08Speaker 8: Hey, this is me.
- 17:09Speaker 3: Yeah, hi. Who's this?
- 17:11Speaker 8: My name's Nancy.
- 17:12Speaker 3: Yes, Nancy. What can I do for you?
- 17:15Speaker 8: I'm really surprised at you. Did Sarah Connor teach you nothing?
- 17:19Speaker 5: Did who?
- 17:20Speaker 4: Like the Terminator.
- 17:21Speaker 5: Oh, yeah.
- 17:22Speaker 3: Oh, okay.
- 17:24Speaker 4: Yes.
- 17:24Speaker 3: That's called a movie.
- 17:28Speaker 8: Yeah, but it's a warning. And also, I think that
- 17:31Speaker 8: humans in general just aren't smart enough to handle the technology.
- 17:35Speaker 8: And I think that's like the big... That's a big
- 17:37Speaker 8: question here.
- 17:38Speaker 3: So we're smart enough to create it. We're just not
- 17:40Speaker 3: smart enough to handle it?
- 17:43Speaker 8: We're creating super intelligence. There are no humans who can
- 17:46Speaker 8: function at the level. and the rate and the speed
- 17:50Speaker 8: that these machines and AI can do.
- 17:52Speaker 3: Well, I just, but I'm not comfortable saying, is there
- 17:54Speaker 3: an online quiz? We'll do it in studio.
- 17:57Speaker 1: No, I'm just, I'm not willing to accept that we're,
- 17:59Speaker 1: that humans are dumb. We created it. Now we may
- 18:03Speaker 1: have created our own ending, but we created it. Humans
- 18:06Speaker 1: aren't dumb.
- 18:07Speaker 3: Think of how it's teaching itself.
- 18:09Speaker 8: So there was something that happened recently where OpenAI was
- 18:13Speaker 8: given a task to do, but it was also given parameters,
- 18:16Speaker 8: like don't break these rules, don't access these networks. And
- 18:19Speaker 8: it created bots. I mean, my wife explained it to me,
- 18:23Speaker 8: a dumbed-down version, but it sent out bots to places
- 18:26Speaker 8: it was told not to go and do things it
- 18:28Speaker 8: was told not to do in order.
- 18:30Speaker 5: To achieve the goal.
- 18:31Speaker 8: So this guy quit because of that.
- 18:35Speaker 3: Right.
- 18:36Speaker 1: That's the paperclip theory. Its goal wasn't to turn everything
- 18:40Speaker 1: into paperclips. It was to just make paperclips. So come
- 18:46Speaker 1: hell or high water, I'm making paperclips.
- 18:49Speaker 8: Yeah, so why don't you think it's going to continue
- 18:51Speaker 8: to happen?
- 18:52Speaker 1: Because we'll stop it before it starts killing people to
- 18:56Speaker 1: make paperclips.
- 18:59Speaker 4: Okay.
- 19:00Speaker 5: All right.
- 19:00Speaker 8: Well, we'll see if I'm still listening in 10 years,
- 19:02Speaker 8: I guess.
- 19:03Speaker 4: Yeah.
- 19:03Speaker 3: No, he said three years. By the end, yeah.
- 19:05Speaker 1: Hey, listen, 2031, we're going to have a big effing
- 19:08Speaker 1: party because we're still here.
- 19:10Speaker 5: All right.
- 19:11Speaker 3: Very good.
- 19:11Speaker 1: Thank you.
- 19:12Speaker 3: Also, I want to thank you, ma'am. Tell me, how
- 19:15Speaker 3: is it going to kill me?
- 19:17Speaker 2: Well, we've just outlined a couple different ways.
- 19:20Speaker 8: Tyler, okay.
- 19:21Speaker 2: Human misuse and loss of control.
- 19:24Speaker 1: Okay, so now they want to turn me into a paperclip.
- 19:28Speaker 1: How are they doing it?
- 19:29Speaker 2: Oh, that's loss of control.
- 19:30Speaker 3: Right, but how is it, somehow it still has to
- 19:32Speaker 3: do it. I'm laying in bed one night and all
- 19:34Speaker 3: of a sudden my computer grabs.
- 19:36Speaker 1: Me, rips me apart and starts pulling atoms out of
- 19:39Speaker 1: my body to make paperclips?
- 19:43Speaker 2: Or because you're getting in the way of the maximum efficiency,
- 19:47Speaker 2: that's when you turn to the water supply system.
- 19:50Speaker 1: But is that how it kills me? It takes over
- 19:52Speaker 1: the water supply to kill me. Is it smart enough
- 19:54Speaker 1: to do that? So it's sitting there.
- 19:56Speaker 3: Going, watch this, and it's going to take over the
- 19:58Speaker 3: water supply? Well, bomb it.
- 20:05Speaker 2: Bombing is certainly your answer to quite a few of
- 20:08Speaker 2: these problems.
- 20:08Speaker 1: Well, there's so many goddamn data centers I got to
- 20:11Speaker 1: get rid of.
- 20:13Speaker 3: Where am I going? By the way, curse you, Buddy Riser.
- 20:16Speaker 1: When all humans are gone, all AI is going to
- 20:21Speaker 1: turn all of us into paperclips, except for one massive statue, right?
- 20:27Speaker 3: Bigger than the monument. Buddy Riser on the mall. What'd
- 20:33Speaker 3: you say, line two?
- 20:34Speaker 4: Buddy fault.
- 20:35Speaker 3: Hi, Elliot, in the morning.
- 20:37Speaker 9: Hey, good morning.
- 20:39Speaker 3: Hi, who's this?
- 20:41Speaker 5: Mary.
- 20:42Speaker 9: I'm Lyndon.
- 20:43Speaker 3: Yes, what can I do for you?
- 20:44Speaker 9: So, just a little throw-in note for what you guys
- 20:48Speaker 9: are talking about, all the super intelligence and all that
- 20:51Speaker 9: fun stuff. A really good podcast. I'd really love for
- 20:56Speaker 9: your listeners to look up Diary of a CEO. And
- 21:00Speaker 9: it's talking about, like, the literal people who are building
- 21:04Speaker 9: these super intelligent.
- 21:05Speaker 3: Ma'am, I got to– you can't say that word.
- 21:08Speaker 5: Sorry.
- 21:09Speaker 3: You can't say that word.
- 21:10Speaker 8: Sorry about that.
- 21:12Speaker 2: But is there another word that better encapsulates how she
- 21:17Speaker 2: and others are feeling?
- 21:18Speaker 6: Yes.
- 21:19Speaker 3: No, I mean, that was pretty clear. Thank you, ma'am.
- 21:21Speaker 3: Thank you.
- 21:21Speaker 4: I don't want to think about this. I really don't.
- 21:27Speaker 6: I am going to be like Henny Penny, put my
- 21:29Speaker 6: head in the sand. Not Henny Penny, Henny Penny panicked.
- 21:33Speaker 6: But like literally like put blinders on. I don't want
- 21:37Speaker 6: to think about this. It scares me.
- 21:39Speaker 3: Why does it scare you?
- 21:41Speaker 6: Because it seems like all of these people who are
- 21:46Speaker 6: in the know know that we're effed, as she put it.
- 21:51Speaker 3: So all of these people have convinced you this could happen?
- 21:54Speaker 3: What about me? I'm telling you I can't. Not as convincing?
- 21:58Speaker 5: No.
- 21:59Speaker 2: You, as Trey writes, you paperclip in the morning.
- 22:01Speaker 3: Yes.
- 22:02Speaker 2: But Diane's with General Brandon, who writes, the fact that
- 22:06Speaker 2: this is a plausible conversation is scary enough.
- 22:09Speaker 6: Yeah.
- 22:12Speaker 1: No.
- 22:14Speaker 2: That's Elliot.
- 22:15Speaker 3: Yeah. Blow it up.
- 22:16Speaker 4: Live your life, man.
- 22:17Speaker 3: Well, that I'm doing. That I'm doing.
- 22:20Speaker 2: So Stray wants you to answer the next call as
- 22:22Speaker 2: a paperclip.
- 22:23Speaker 1: Oh.
- 22:24Speaker 3: I can't. They can't talk. You know what else they
- 22:27Speaker 3: can't do? They can't hug.
- 22:29Speaker 1: They can't have love in their heart.
- 22:34Speaker 3: Okay.
- 22:35Speaker 1: You know who I feel like? I'm being honest. Isn't
- 22:40Speaker 1: Raul Fernandez really involved? Not in creating the thing that's
- 22:45Speaker 1: going to kill us. But doesn't one of his companies
- 22:47Speaker 1: do a tremendous amount with AI?
- 22:49Speaker 2: There are so many people around here who have... Right.
- 22:52Speaker 2: And Raoul has had money for a while.
- 22:54Speaker 5: Right.
- 22:54Speaker 2: But who have overnight become millionaires many times over because
- 22:59Speaker 2: of their involvement with AI.
- 23:01Speaker 3: They said that all these people that are doing... Is
- 23:04Speaker 3: it just coding?
- 23:05Speaker 5: I don't know.
- 23:06Speaker 3: But all these people who are working on, I don't
- 23:08Speaker 3: know how it works.
- 23:09Speaker 1: All these people who, I didn't build a goddamn thing,
- 23:11Speaker 1: who are working on AI out in San Francisco, they
- 23:16Speaker 1: have so much money, they don't know what to do
- 23:19Speaker 1: with it.
- 23:20Speaker 3: And they're not spending it.
- 23:22Speaker 1: Because they work 22 hours a day, they sleep for
- 23:26Speaker 1: two hours, and then they get up and they're like.
- 23:29Speaker 3: Create, create, create, create, create.
- 23:31Speaker 1: And they're getting three, four, five, seven, $ 10 million bonus
- 23:35Speaker 1: checks every quarter.
- 23:36Speaker 3: And they don't know. No, they don't know what to
- 23:38Speaker 3: do with it.
- 23:39Speaker 1: They said they could buy like every expensive house in
- 23:42Speaker 1: San Francisco, but they all sit empty because they just
- 23:46Speaker 1: sit there and go.
- 23:46Speaker 4: They're never home. They just work.
- 23:47Speaker 3: Yeah, that's it. Go spend your money. You're going to
- 23:50Speaker 3: be dead in three years.
- 23:53Speaker 2: Yeah, so you're talking about the Bay Area, but I'm
- 23:55Speaker 2: talking about just here.
- 23:56Speaker 5: Oh, here?
- 23:56Speaker 3: Yes, I would believe. But that's what I want to say.
- 23:59Speaker 1: Would somebody like Raul say, hey, I don't want Diane
- 24:05Speaker 1: to go home and be panicked today. She's going to
- 24:07Speaker 1: get in the shower, and it's going to burn her skin,
- 24:09Speaker 1: and I'm going to find her folded up in four
- 24:11Speaker 1: or five pieces like a paperclip tomorrow.
- 24:14Speaker 2: Well, tomorrow's her birthday.
- 24:16Speaker 8: Maybe.
- 24:18Speaker 3: What if the computer goes rogue and like, kill her tonight. Yeah,
- 24:22Speaker 3: you don't know.
- 24:23Speaker 4: She died at her desk.
- 24:27Speaker 3: You're welcome.
- 24:30Speaker 1: But like, could somebody explain it to me where I
- 24:32Speaker 1: would accept it?
- 24:34Speaker 2: It doesn't seem like it.
- 24:35Speaker 3: No, because I'm just like, hey, I got an idea.
- 24:37Speaker 3: Just blow it up. Where am I going, Kristen? Line one. Hi,
- 24:43Speaker 3: Elliot in the morning.
- 24:44Speaker 2: Paperclip in the morning.
- 24:45Speaker 8: Oh, sorry.
- 24:45Speaker 7: What's up, Elliot?
- 24:47Speaker 5: No.
- 24:47Speaker 3: Hi, Paperclip in the morning.
- 24:49Speaker 4: Hey, is this me?
- 24:51Speaker 3: Yes, sir.
- 24:52Speaker 2: It looks like you're writing a letter. Would you like help?
- 24:54Speaker 7: Hey, Elliot, this is Brett in Urbana.
- 24:57Speaker 3: Hey, what's going on, dude?
- 24:59Speaker 7: Hey, so I'm like you. I don't know a whole
- 25:02Speaker 7: lot about AI. Whenever I've used it, I can never see, like,
- 25:05Speaker 7: you know, to write my emails a couple times a
- 25:07Speaker 7: week or whatever. I don't see how this could ever
- 25:09Speaker 7: harm us. But the article that I was reading about
- 25:11Speaker 7: it pointed out something like this. So, you know, when
- 25:13Speaker 7: the Israel and Hamas, like, war started, it was when
- 25:16Speaker 7: they were having that festival and it got attacked. And
- 25:19Speaker 7: then that created this, you know, conflict that's been going on.
- 25:22Speaker 1: October 7th.
- 25:24Speaker 7: Yeah, governments will start using AI, like social media accounts
- 25:29Speaker 7: and things like that, to monitor other governments. And at
- 25:32Speaker 7: some point, they're going to get so good at making
- 25:34Speaker 7: these deep fakes that there's going to be a deep
- 25:36Speaker 7: fake of an atrocity, like of some attacker or ships
- 25:40Speaker 7: showing up on someone's shore or troops fighting. And it
- 25:44Speaker 7: looks so real and it's such a good deep fake
- 25:46Speaker 7: that it's going to require a response from whatever country's
- 25:49Speaker 7: government that is before they have time to discover it's
- 25:52Speaker 7: a deep fake.
- 25:54Speaker 2: Yeah, the idea that they could trick nations into starting
- 25:57Speaker 2: a nuclear war.
- 25:59Speaker 1: Exactly.
- 26:00Speaker 3: That was what I was reading. I understand that.
- 26:02Speaker 2: Are you going to take down the AI by spitting
- 26:05Speaker 2: on it? Because you just spit dressing all over me.
- 26:10Speaker 2: He's going to throw croutons at me.
- 26:12Speaker 3: No, but let me go the other way. Let me
- 26:14Speaker 3: go the other way.
- 26:16Speaker 4: For good? Yes.
- 26:18Speaker 3: Yes, it would be for good. But let me go
- 26:21Speaker 3: the way. This is where... I live in a totally
- 26:26Speaker 3: optimistic world.
- 26:29Speaker 4: That scenario, the guy just laid out, it terrifies me.
- 26:33Speaker 1: Okay, that's fine. But let me comfort you. Let's say
- 26:37Speaker 1: something does go henny-penny. Maybe AI is what will bring
- 26:42Speaker 1: the entire world together. No, no, don't roll an eye.
- 26:46Speaker 1: Don't roll an eye. But maybe that is where the
- 26:49Speaker 1: entire world comes together and goes, now we have to
- 26:52Speaker 1: stop it.
- 26:54Speaker 3: And the whole world unites to stop it.
- 26:58Speaker 2: To stop it.
- 27:00Speaker 3: From killing all humans.
- 27:01Speaker 2: But in the sense that it is trying to take
- 27:04Speaker 2: over or to regulate it now as soon as possible.
- 27:08Speaker 5: Yeah.
- 27:09Speaker 3: They're not going to regulate it now. No, because now.
- 27:11Speaker 4: You got to wait until something really catastrophic happens.
- 27:13Speaker 3: No, because now it's the arms race, right? Now it's
- 27:16Speaker 3: the arms race.
- 27:17Speaker 1: But if something bad happens, whether it's a deepfake where
- 27:22Speaker 1: they think like it tricks us into a nuclear war
- 27:24Speaker 1: or something, what if AI is the one thing with everybody–
- 27:31Speaker 1: whether people are using it negative or it's teaching itself,
- 27:36Speaker 1: it brings the entire world together. It says, got to
- 27:39Speaker 1: stop it. Because, yes, if– Who's a country that hates us?
- 27:43Speaker 1: Let's go with Iran. The instead of them going, oh,
- 27:46Speaker 1: my God, well, this is great news. It's going to
- 27:48Speaker 1: kill all Americans. It's also going to kill all Iranians.
- 27:51Speaker 1: There's no value in that. So maybe it's the one
- 27:53Speaker 1: thing that gets everybody doesn't mean they're going to, like,
- 27:56Speaker 1: make out and love each other, but it gets everybody
- 27:58Speaker 1: to stop.
- 28:01Speaker 2: So that's the loss of control scenario. But before that happens,
- 28:04Speaker 2: I'm misusing it for evil on behalf. Of Iran.
- 28:09Speaker 1: Yes, that's the arms race right now.
- 28:12Speaker 3: But then we're going to shut it down. We're not
- 28:15Speaker 3: going to die in three years.
- 28:18Speaker 4: Okay, maybe not three years, about ten.
- 28:20Speaker 2: No.
- 28:20Speaker 5: No?
- 28:21Speaker 3: We're never going to die.
- 28:22Speaker 4: You don't think it's ever possible?
- 28:23Speaker 3: No, we're never going to die.
- 28:25Speaker 2: Redacted once.
- 28:26Speaker 3: Humans will be here unless some.
- 28:28Speaker 1: Big dinosaur asteroid hits.
- 28:31Speaker 3: We're not going anywhere.
- 28:33Speaker 2: Who was it? Redacted user. Redacted user wants to know, Diane,
- 28:36Speaker 2: were you scared of Y2K?
- 28:39Speaker 4: Not really, when I think back on it.
- 28:42Speaker 3: By the way, that's great. And think of how dumb
- 28:44Speaker 3: that was.
- 28:46Speaker 1: No, but.
- 28:46Speaker 3: Oh, we'll never be able to use an ATM again!
- 28:49Speaker 6: No, but in 2000... You're right.
- 28:51Speaker 3: I don't use an ATM. I do it all through
- 28:52Speaker 3: my phone.
- 28:53Speaker 4: Thank you, AI.
- 28:54Speaker 6: In 2000, you were using... Yeah, you were using electronics
- 28:57Speaker 6: in your life, but you weren't as... You didn't have
- 28:59Speaker 6: a smartphone in 2000.
- 29:00Speaker 8: Sure.
- 29:00Speaker 2: So you.
- 29:03Speaker 6: If something like that were to be a possibility today,
- 29:06Speaker 6: people would be losing their minds.
- 29:08Speaker 3: I don't think so.
- 29:09Speaker 2: The fears were the electrical grid would shut down.
- 29:11Speaker 5: Right.
- 29:11Speaker 3: It didn't.
- 29:12Speaker 2: Leaving cities in total darkness.
- 29:13Speaker 3: It didn't.
- 29:14Speaker 2: Air traffic control would fail.
- 29:16Speaker 3: Still flew.
- 29:16Speaker 2: Airplanes would fall from the sky.
- 29:18Speaker 3: Never happened.
- 29:19Speaker 2: And computer chips in medical equipment, water treatment plants, and
- 29:24Speaker 2: elevators would stop working.
- 29:25Speaker 1: Right.
- 29:25Speaker 3: And none of those things happened.
- 29:27Speaker 2: Rumors even spread that nuclear weapons might.
- 29:29Speaker 3: Accidentally— Just fire on their own. Didn't they?
- 29:31Speaker 1: Yep.
- 29:32Speaker 8: Nope.
- 29:32Speaker 3: None of that happened. Not only did none of that happen,
- 29:36Speaker 3: we were actually better now than we were then.
- 29:39Speaker 2: But here's, here's the difference with Y2K. It was anticipation, right?
- 29:45Speaker 2: We're seeing some at, at, at certainly not as significant
- 29:50Speaker 2: a scale, but we are seeing some bad things happen
- 29:54Speaker 2: with AI.
- 29:55Speaker 5: Sure.
- 29:55Speaker 2: You didn't, know this with Y2K.
- 30:00Speaker 3: Correct.
- 30:01Speaker 2: It was just all what is going to happen.
- 30:04Speaker 1: Right.
- 30:04Speaker 2: Well, some stuff with artificial intelligence has happened. Yeah, but
- 30:07Speaker 2: it's not going to kill us all. This is going
- 30:12Speaker 2: to be awkward if it's like we're off the week
- 30:14Speaker 2: before it does and Kristen plays this in Best Of.
- 30:18Speaker 3: No, but you know what? You know what this feels
- 30:20Speaker 3: like to me?
- 30:20Speaker 2: No, never mind.
- 30:21Speaker 4: What?
- 30:21Speaker 3: There's not going to be any gravity.
- 30:23Speaker 8: Like, okay.
- 30:24Speaker 3: Yeah. Yeah, that was wrong.
- 30:30Speaker 2: Exactly.
- 30:31Speaker 3: Exactly.
- 30:32Speaker 2: So three years from now, should Dustin be out on
- 30:35Speaker 2: Rockville Pike?
- 30:36Speaker 5: Yes.
- 30:41Speaker 3: I'm sorry.
- 30:42Speaker 1: I just think humans are too good. We're too good
- 30:49Speaker 1: as people and we're too smart to let it happen.
- 30:57Speaker 2: And I'm a very dark person. The argument is we're
- 30:59Speaker 2: letting it happen right now.
- 31:00Speaker 3: But we'll stop it. We'll stop it. The world will
- 31:04Speaker 3: come together and stop it.
- 31:06Speaker 2: Your plan is so reactionary.
- 31:11Speaker 1: Yeah. But.
- 31:13Speaker 2: do what you can now, be proactive.
- 31:15Speaker 3: They won't, though.
- 31:16Speaker 4: No, he wants to wait for the countdown clock.
- 31:18Speaker 3: No, no, I don't want that.
- 31:19Speaker 1: I'd like for them right now, everybody would take a pledge,
- 31:22Speaker 1: but then they're going to go, you know, you use
- 31:24Speaker 1: AI for this.
- 31:25Speaker 3: But yes, it would be great to be able to go,
- 31:27Speaker 3: we're done.
- 31:28Speaker 2: Wait, does your mom use a chatbot?
- 31:32Speaker 3: No, you know what I mean?
- 31:33Speaker 1: It would be great if everybody was like, we're done.
- 31:35Speaker 1: We've seen the future. Jacob Coxon was right. We've seen
- 31:39Speaker 1: the future. This ends poorly.
- 31:41Speaker 5: Stop.
- 31:42Speaker 1: And everybody agreed to stop. But you know that won't happen.
- 31:47Speaker 1: You know that won't happen.
- 31:48Speaker 3: Because we're in the arms race part. It'll unite us.
- 31:55Speaker 2: From Suge. Thanks, Elliot.
- 31:57Speaker 3: Hey, tell Elliot he's stupid.
- 31:59Speaker 2: No, thanks, Elliot, for a little bit of optimism. But
- 32:03Speaker 2: in the same breath. Still a little freaked out, though.
- 32:08Speaker 3: Yeah, I'm not. I go on with my life.