Yes, it is a generalization, but one that I also believe to be fairly universally applicable in this case. There is a certain philosophical underpinning here - essentially to believe is “machine intelligence” one must internal debase their assessment of human intelligence. I do not believe is “machine intelligence” and “artificial intelligence” is an oxymoron. The map, not matter how rich and detailed and immersive, can never be territory. Similarly the tool is not the talent. Saying “the machine is so smart” is like saying “this calculator is good at Math” or “my POSCA paint pens are really good at art.” The machine or tool is an inert object outside of the human operator. As to what nebulous and generalized “existential dangers” that AI may pose (note, AI people never make this specific, always generalized), perhaps there is not much that can be done other than actively opting out of engagement with these systems when you can. Thats what the individual can do. Beyond that, I don’t think of “what should be done” (ie appealing to be centralized decision authority - as this the same thinkings that leveraged the fear of nuclear weapons to profane the blessing of nuclear power while only INCREASING the number of nuclear weapons in play) - but I do think it terms of what I think will happen. And what I think will happen is this. OpenAI and Anthropic are AI service provider / hardware buyer duopolies that rest atop Nvidia as the hardware producer monopoly. OpenAI and Anthropic are perpetually insolvent and have no path to profitability - they are capital destruction machines fueled by fiat VC slush. Eventually the bill will come due and all services that are dependent on OpenAI or Anthropic will either be squeezed to death by cost increases or while die when OpenAI and Anthropic collapse financially. They may be bailed out by the government, but this will truly set us on the path we of realtime USD currency debasement, with the government bankrupt itself entirely. SO, I think AI can only be a useful technology if it’s true costs are borne by the market around it. After the collapse of OpenAI and Anthropic (and general stock market value destruction), the costs of AI will not longer have “free money” lined up to paper over. Individuals and companies will have to operate their AI decentralized at the compute level. And this recognition of costs will ensure that AI, where is it used, will be used in a way that adds value for humanity. Short of this AI is just another fiat malinvestment. The tool itself and fear of the tool will both be leveraged by people who bear not actual costs to attempt to lock in their control over others who are too lazy or immiserated to understand that they can opt out (though opting out has its own costs). They don’t actually have the resources to achieve this…unless of course they can conquer your mind first, as this is the ultimate force multiplier.