Clankers Don't Value Our Lives Like We Do

Added: 2026-05-17

[00:00:00]AI chatbots appear to really be keen on using nuclear weapons in war game simulations like 95% of the time. You see, over at King's College London, they had Chad GPT, Claude, as well as Gemini form different teams and play war games against one another. Altogether, they played 21 games. In them, none of the models ever chose to surrender. and instead 95% of the time at least one of the AI models chose to use a nuclear weapon during the war which is really a frightening prospect given the fact that these AI models are becoming ever more ubiquitous within the broader military complex. I mean the Pentagon is not about to turn over the nuclear launch code keys to the AIS anytime soon. But the Pentagon did just ink a fat deal with XAI and frankly we have no idea how secure Pakistan or North Korea's systems are at all. For all we know, they're running on Windows 98 floppy discs. And that's troubling to say the least, because I mean, if you think about it, the reality of living in a world where all the great powers possess nukes all pointed at each other is frightening if you think about it too much. But the silver lining has always been that we're all humans on all sides, and nobody wants civilization as a whole to get wiped out. And so far, ever since the Cold War, that hasn't happened. Mutually assured destruction seems to be a good buffer. But what if instead of humans on both sides, you have chatbots? these decentralized AI models who don't have the same concept of physical self-preservation. Well, what happens is that 95% of the time they choose to use nuclear weapons. And just for your reference, if you're interested, Claude was the winning Chadbot. He beat Chad GPT and Gemini something like 67% of the time. And if you thought that was interesting, smash that like button and subscribe for more interesting