Siliconsciousness: With AI, are We Walking the Dog or is the Dog Walking Us?
The DSR Network
This episode explores the growing crisis of autonomous AI management as host David Rothkopf and guest Camille Stewart Gloster examine why voluntary industry self-regulation
Key takeaways
- Camille Stewart Gloster argues that voluntary self-regulation has failed to provide meaningful guardrails, asserting that government must establish clear red lines and operational guidelines for autonomous systems.
- The host introduces a 'walking the dog' analogy to illustrate how organizations often grant AI agents excessive permissions without understanding their emergent behaviors or potential for harm.
Main topics
- Autonomous AI governance
- Corporate insider risk management
Notable quotes
"It is not their job to self-regulate. It is the job of government to impose restrictions and guidelines and at very least the red lines under which they should operate."
Conclusion
The episode underscores that harnessing autonomous AI requires discipl
Transcript preview
Speaker 1 (0:01) Welcome to Scylla Consciousness, the DSR Network podcast focusing on the artificial intelligence revolution, politics and policy. Speaker 4 (0:19) Hello and welcome to Scylla Consciousness. I'm David Rothkopf, your host, and this week we're very glad to be joined by a friend of ours for two reasons. Camille Stewart Gloucester is CEO of CAS Strategies and previously served as Deputy National Cyber Director for Technology and Ecosystem Security at the White House. The other reason is she's got a new book, and the book is The Insider You Built, How Organizations Stay in Control of Autonomous AI Agents. Congratulations, Camille. Speaker 2 (0:55) Thank you so much, and it's good to be back. Speaker 4 (0:57) It's good to see you. It's good to talk to you. And lest anybody doubt. Speaker 4 (1:05) your mastery of this issue, the fact that you somehow got the President of the United States, who's often asleep, but somehow got the President of the United States to say, you know what the most important issue is in the world right now? Managing autonomous systems. And I'm going to put together a brains trust of people to be in charge of it. Elon Musk, okay? And Newt Gingrich. That's kind of weird. And what is this other dude, Lucky, who's a business partner with Don Jr.? And I thought, wait a minute. I know somebody that should have been on this group. Not that you would have fit right in. Speaker 2 (1:46) Yeah. I Speaker 4 (1:48) don't know. But it does seem like this is front of mind for everybody, although. I'm not sure if your focus would be how to get maximum lethality from autonomous weapons platforms without weirdos, beardos, and the other things they don't want involved. It's a strange time, but let's start with the news. What did you think of that development from the Pentagon? Speaker 2 (2:20) You know, there's so much. Speaker 2 (2:24) swirl, but not a lot of substance happening right now. We've got voluntary agreements to self-regulate. We've got committees forming. Hey, Speaker 4 (2:32) hey, the president said their voluntary agreement was like the constitution. Speaker 2 (2:38) Yeah. I mean, it's morally binding. It's Speaker 4 (2:40) morally binding. What else could you ask for? Exactly right. Speaker 2 (2:46) I mean, it's great to see it getting a lot of attention. As far as responding to even the organization's requests for more guidance on how to operate in this moment or to coalesce around a slowdown, it has not met the mark in that way. But I do hope by some... we start to move in the right direction. I mean, unwittingly, all of these companies are at the front lines of writing policy, of shaping society and democracy, and they were not made to do so. So hopefully something changes. Speaker 4 (3:22) Yeah. Okay. When you say they're not made to do so, I got a problem with the whole formulation. I got a few problems, but one of the problems is Speaker 4 (3:33) you know, who's stopping them? Well, that's part of the problem. Right. But the point is, if they want to have guardrails, they could have guardrails. If they didn't want to develop systems that would destroy the world, Speaker 1 (3:46) they Speaker 4 (3:47) don't have to do it. You got to sort of say, what's their real agenda here? Because, you know, stop me before I kill again is not super persuasive. Speaker 2 (3:58) Not at all. I mean, honestly, that's where government is supposed to stand in, to shift the market towards what is right, to highlight what is in the best interest of the citizens and the society they govern. And the job of businesses is to make money. They are doing what will help them make money. I mean, fundamentally, even if as individuals, some of them care about what is happening to society and democracy, their job is to... They're beholden to their shareholders and their job is to make money. So it is not their job to self-regulate. It is the job of government to impose restrictions and guidelines and at very least the red lines under which they should operate. Speaker 4 (4:45) Right. But so why then, you know, if you're the CEO of one of these companies, do you say, we could destroy the world and we don't think the government should step in and we'll handle it from here? Even though they could have been handling it from here all along. Well, that's the thing. Speaker 2 (5:02) They could have been handling it all along. What I do think is emerging to the top is the fact that being thoughtful about governance and security and investing and actually controlling the environment that they're creating is actually good business sense. And we should have known that from the beginning, but they didn't. Moving fast sounded... far better and wanting to have the best capability. But quite frankly, if nobody trusts the capability, if it's hacking government websites and other organizations, and you had no idea these things were happening, I mean, there's a real, at the very least, PR problem and likely to build into other problems. And it's definitely a national security, economic security risk. And so it's become very clear, painfully clear, almost, that governance and security and investing in actually understanding and controlling the technology we're letting loose into your environments and into the world is a business imperative. Speaker 4 (6:04) Yeah, and that's, I think you put your finger on it there, because they have taken the position of go fast, we're going to change the world, because that's what built multiples based on projected future earnings. And now they see that there is some prospect that those earnings could be upset by incidents where they're out of control and they're like, hmm, this could be a problem. But everything they do, particularly the broligarchy that Trump had to the sheet dinner and then had with him a day or two later. Everything that they do is oriented towards share price. Now, you might say, well, that's why they're there. But on the other hand, when you're dealing with, I don't know, the fate of humanity, perhaps something else ought to figure in. And, you know, I think with this book, you're putting your finger on one of the central issues. I mean, as you know, there's new polling out today on how uncomfortable Americans are with AI, much more uncomfortable. than people in other countries, as it happens. And interestingly, young people who have more exposure to AI are more uncomfortable with it. And I think there are a couple reasons for that. One of the reasons is the Brologarks, right? One of them is that somehow they've put together this kind of rogues gallery of the weirdest, most egotistical, fucked up, ketamine-taking... Speaker 4