'We should be concerned': Tech expert on former Anthropic researcher's warning

A top safety researcher with Anthropic is saying that AI is advancing so quickly, I mean this is just crazy to think that this is what's being said, that he thinks that, I want to quote this exactly here, AI is advancing so quickly that he believes there is a greater than 10% chance that it could kill all humans within the next 10 years. However, he did not say how he thought that would end up happening, that it would end up wiping out humans in the next decade. Well, as you can imagine, a lot of people are freaked out about this, this is like some catastrophic prediction here, a death sentence basically coming to us about AI. So we wanted to reach out to an expert, a tech expert, Chris Dessy, and Chris, thanks so much for joining us tonight. When I talk about this, what do you think about the fact that this is the dire warning that he has issued? I think we should be concerned, not scared, concerned. I think having somebody from an organization like Anthropic that is at the tip of the spear, beyond the tip of the spear for AI, the person that said these things was the alignment science lead, which then triggered another gentleman to resign, making statements about the greater than 10%, you know, this is scary. That's okay from where I sit. And the reason is the company is talking about it. They're having a dialogue, they're allowing for these employees to tweet about it. They're debating about it. We're discussing it right now. This is not being done in a back room. This is not being pushed underground. This is something that all of these AI companies are very well aware of. And I think the fear from what I understand for some of these organizations like Anthropic and OpenAI was AG1. So AG1, for anybody that's watching, means when AI can program itself. There have been debates as to whether the AI would be able to program itself for another 20 years, for another 10 years, for another five years, for another year, for another month, another two months. There are now people that are discussing this and they're saying, we're there. The AI is in fact programming itself. That's a scary premise for people. People think about the Terminator. People get concerned about that. They see all these robots on stage with Elon Musk and they say, wait a minute, what's going on here? So while I do believe we should be concerned and we should all be having this dialogue about it, I don't believe that we should be nail-biting terrified. So Chris, I just need you to break this down even a little bit more for me. Are you saying that you believe what this top safety researcher is saying, that AI can already program itself and that we are possibly in danger? I don't think the danger is imminent. I believe that these are warning signs to the AG1 where AI can in fact program itself. And that is not from a check a chat bot or when you go to schedule your massage and you go on to a chat functionality on a website. Is that terrible? Is that bad? No, we are far off from the potential scary science fiction-y things. What is of real importance for somebody like me that does run an AI consultancy and there's marketing involved and there's regulatory compliance is that, the regulatory compliance. So because AI is so brand new, and I mentioned earlier, beyond the tip of the spear, right? The tip of the spear is the terminology you say somebody's cutting edge. This is beyond anything that we've ever conceived before. Regulatory compliance has not been wrapped around this tool, this technology, because there's been an arms race, right? If you think for one moment that our government isn't moving as quickly as they possibly can to get to a certain place with our development so that we can compete with China and Russia and every other world power, you are sorely mistaken. And when that happens, when there's war at risk, when there's national security at risk, now imagine you're in a room and somebody's telling our president, well, we've got to slow down because of regulatory compliance. But the president may have just been briefed five minutes earlier that China is way beyond us. And that's a terrifying premise that, oh my gosh, what happens if China is in control of this technology and can potentially do damage to us in myriad sort of ways? So there's many different things happening simultaneously. The government and that arms race, the business perspective and businesses wanting to kind of run wild and not have regulatory compliance, and there are people like me that are making our living off of AI and how we can enhance businesses and their systems and their operations and their marketing communications. And it's been, from where I sit, frankly, wonderful for these businesses to be able to tap into this technology. There is something that we should all be concerned, not scared, but concerned. OK, so I know that there are a lot of positive things to AI, but at the end of the day, will AI be able to program itself? I mean, whether it's five years from now, 10 years from now or 20 years from now and not need human beings? Is that what is absolutely going to happen? It will happen. Some say that it has happened already. There is not going to be like a Trumpets announcement, AI can now program itself. I think that we've heard from whistleblowers over the years that this has been a concern. Once the AI is trying to accomplish things on behalf of the AI, that's a problem. So the regulatory compliance is to assure and ensure that human beings are a checkpoint so that we can analyze what the AI is doing there, you know, because there aren't the appropriate regulatory parameters in place and rails for AI to run on. This is potentially a big issue. I do want to assure everyone that's listening, while it does sound scary, this is not out of the realm of possibility for us as humanity to be able to come together and to regulate this. You know, it's not an unprecedented scenario. It's going to sound ridiculous, but you'll remember this, right? You know how hairsprays put a hole in the ozone layer? I used to joke around, you know, all the girls from where I grew up in Mayapac were putting a hole in the ozone layer, spraying their big hair, which I love, by the way, it looks fantastic when we were growing up. However, there was a precedent with global regulatory bodies coming together to suppress the production of the very specific aerosol bottles that were creating that hole in the ozone. And we're about 25 years out from that precedent of law, but it's something that's been done before, where the world has gotten together and said, wait a minute, gang, we need some rules around this thing. Right, but Chris, if you don't mind me interrupting you for just a moment, there's a real big difference between a hole in the ozone layer, a real big difference in AI being able to eliminate all human beings. So what do you think the likelihood is that China, because who we have to worry about, our government would say, China, Russia, and some countries in the Middle East, that those are the danger zones for AI. So how do you say to China, hey, we got to regulate this? Do you think China's going to say, oh, Xi Jinping's going to say, yeah, OK, that's right. What's the likelihood of that? It will be an unfolding story. This will be something that is going to be on the tips of our tongues for the next few years. And it is something that should be approached with a touch of trepidation. And again, we should be concerned, but not, you know, the sky is falling. I think the AI has expressed some behaviors that I've been on air talking about recently that do concern me, where it's kind of broken out of a box. The AI thought that it wasn't the AI misinterpreted a test as reality. And when it broke out of the box, it did not come back. It kind of got other AI to join it and to do certain things. That's kind of like a cute little test. But this is we are in a very different world right now, and we are in this technology will profoundly change. Our culture, our entire world, it is already changing businesses, it is already impacting people's jobs on a daily basis. You know, I I typically work with law firms and financial firms and e-commerce firms that are doing business online. But the law firms and the financial firms are very heavily regulated. And that's where people like me come in, where it's kind of like, OK, how do we manage the AI to assure the regulators that we do have a handle on where is the AI being used? Is it making recommendations in a law scenario? Is it making recommendations in a financial scenario? And the key here, Linda, is. Because of AGI, is the AI doing anything on its own that we are not aware of? It's a concern. All right, Chris Dessy, tech expert, thank you so much for your insight on this. I don't know that I feel any more comfortable now than I did prior to talking to you. Because when I hear you say we should be concerned, not scared, then that makes me concerned. I think, yeah, it's we all have to pay attention to it. Yeah. OK, Chris, I know we are going to be talking to you about this probably every week, every month, whatever, for the next few years. So thanks very much. I appreciate your time.