AI is already having negative impacts on working families. It’s sucking up a huge portion of the money in the economy, and the money is being spent on things that produce very few jobs (data centers). The stated goal of these companies is to reduce jobs, increase surveillance, and to make the war machine more deadly. Some state there’s a good chance what they’re actively working toward will destroy humanity. Some believe one company will create an ASI, starting the technological singularity, which will concentrate immense power into that single company that beats everyone else to it. Arguably, an AI crash is preferable to an AI “success.”

Same, I don’t find “vibe-coding” very useful. It just produces slop. With enough back and forth with an LLM I can usually get something that “works,” but works poorly. I don’t have the motivation to read the slop it produces, so I don’t know if it’s doing things correctly or not (when I do investigate, it often is not). I often feel like the time I spent prompting the LLM in English would’ve been better spent “prompting” the compiler or interpreter in a language designed for software development.

I once saw an old lecture where the guy working on Yahoo spam filters noticed that spammers would create accounts to mark their own spam messages as not spam (in an attempt to trick the spam filters; I guess a kind of a Sybil attack), and because the way the SPAM filtering models were created and used, it made the SPAM filtering more effective. It’s possible that wider variety of “poisoned” data can actually help improve models.