🚬🐶

Anything branded "AI" is automatically grift-adjacent. "Artificial intelligence" is a non specific term which truly means nothing. It can apply to basic statistical models, massive natural language models, image processing models, to prediction or categorization or input-output models. It's just completely non specific in a way that makes it useless. If something is called "AI" the first question should be "No but what is it actually? What math are you actually doing?"

Chatbot style input-output LLMs appear to be uniquely horrible for the human brain. We are very bad at interacting with something in natural language (even very broken natural language) and not projecting human intelligence on it. This is well know as the ELIZA effect and it appears to be pretty hard coded into human cognition. Humans have a strong tendency to hand over their critical thinking to these tools in a process known as cognitive offloading. This alone makes me believe that chat interfaces should be effectively banned from consumer facing products, and question if they have any use anywhere. It's unclear if this effect is unique to the chat interface or inherent to natural language models of all kinds, but regardless we should not be rolling these tools out in schools until we know.

"AI" companies have essentially exclusively behaved badly. They kicked off their products with the largest scale theft of intellectual property in history, one that's unlikely to ever be prosecuted. They peddle their products as tools to help people cheat, lie, and grift. They openly valorize men who say truly despicable things (one, two, three to start). They're rotting the US economy and betting the entire house on rolling snake eyes. They're using underhanded tactics to get their products into businesses and schools, and to spam their polluting energy hungry data centers across the globe. They're categorically untrustworthy.