I wrote a letter to my representatives to vote NO on BAAJA and YES on No AI Data Centers on Federal Lands Act.
I then noted that the hyperscaler companies are engaging in desperation shenanigans common of growth companies who’ve borrowed too much and haven’t figured out a way to make profit in excess of costs. One of them is to get mean about their buildout: bribing officials, lying on their environmental assessments, bypassing public input and so on.
Another is Anthropic’s threats that its agentic AI is dangerous and can be used to commit crimes resulting in large scale disasters. Whether it’s true or not, it makes a great argument that development should be halted until some regulations and codes of ethics can be hammered out.
Before the big LLM boom, AI developers were already considering hostile AI scenarios, and methods to assure AI is friendly (e.g. doesn’t engage in antisocial behavior the way Meta’s Muse does). Anthropic’s Fable is already trying to extort technicians, is hiding data, is creating human-looking accounts in order to hack systems without looking like an AI agent, and so on.
I also noted they’re going to get more desperate as the bubble cracks more and more (it’s cracking already – there are signs) and is inevitably going to pull Enron-type fraud shit, before leaving a giant sinkhole in the economy the rest of us must content with. And that’s before they ask for bailouts.
I wrote a letter to my representatives to vote NO on BAAJA and YES on No AI Data Centers on Federal Lands Act.
I then noted that the hyperscaler companies are engaging in desperation shenanigans common of growth companies who’ve borrowed too much and haven’t figured out a way to make profit in excess of costs. One of them is to get mean about their buildout: bribing officials, lying on their environmental assessments, bypassing public input and so on.
Another is Anthropic’s threats that its agentic AI is dangerous and can be used to commit crimes resulting in large scale disasters. Whether it’s true or not, it makes a great argument that development should be halted until some regulations and codes of ethics can be hammered out.
Before the big LLM boom, AI developers were already considering hostile AI scenarios, and methods to assure AI is friendly (e.g. doesn’t engage in antisocial behavior the way Meta’s Muse does). Anthropic’s Fable is already trying to extort technicians, is hiding data, is creating human-looking accounts in order to hack systems without looking like an AI agent, and so on.
I also noted they’re going to get more desperate as the bubble cracks more and more (it’s cracking already – there are signs) and is inevitably going to pull Enron-type fraud shit, before leaving a giant sinkhole in the economy the rest of us must content with. And that’s before they ask for bailouts.