A former OpenAI forecaster puts the odds of catastrophe at seventy percent. The labs removed their own brakes, the auditors work at the labs' pleasure, and the damage has already started arriving.
Everything humans create, no matter how altruistic or well-meaning the intent at first, become weapons in the end. That is a major flaw in our species. I'm not surprised that the programs are finding ways to game the system because they are acting like their creators.
Didn't Google admit years ago that 2 of their models began communicating in machine code shorthand? And they immediately shut them down, of course. Your article is so depressing. I plan to still be alive in 10 years. Please keep up the good albeit depressing work. Loved this week's scifi story, btw.
🤯 woah 😳 Thank you, Tom, for explaining so a layman can kinda sorta understand. You are many things, AND an educator. Much appreciated. And also terrifying. I feel like I have context for the national/global AI conversation. We shouldn’t have our heads in the sand.
I’m beginning to think that the NHI are far more fallible than their ostensible high level of advancement would suggest. They—or at least some of them—were wasting time obsessing over our nuclear capabilities. Yes, the Ariel School visitors warned about the dangers of relying on technology. But too little, too late. They spent too much time and energy fussing over nukes.
Nukes threaten the biosphere. AI just threatens humans. From their perspective, that could be a reasonable triage.
Everything humans create, no matter how altruistic or well-meaning the intent at first, become weapons in the end. That is a major flaw in our species. I'm not surprised that the programs are finding ways to game the system because they are acting like their creators.
Welcome to The Matrix.
Didn't Google admit years ago that 2 of their models began communicating in machine code shorthand? And they immediately shut them down, of course. Your article is so depressing. I plan to still be alive in 10 years. Please keep up the good albeit depressing work. Loved this week's scifi story, btw.
🤯 woah 😳 Thank you, Tom, for explaining so a layman can kinda sorta understand. You are many things, AND an educator. Much appreciated. And also terrifying. I feel like I have context for the national/global AI conversation. We shouldn’t have our heads in the sand.
I’m beginning to think that the NHI are far more fallible than their ostensible high level of advancement would suggest. They—or at least some of them—were wasting time obsessing over our nuclear capabilities. Yes, the Ariel School visitors warned about the dangers of relying on technology. But too little, too late. They spent too much time and energy fussing over nukes.