HN user
I see a big difference between AI people choose to use themselves (eg ChatGPT), and AI people are forced to use (whether unhelpful customer-service bots, or LLMs attempting to cover up for poor user interface design).
It is sad that this person got scammed, and there are a lot of legitimate comments in this article. But I have to admit that at first I thought the "scam" was going to be that these $200 smart smoke alarms aren't a real product. Certainly for me, the risks of adding such complexity to such an important product are not worth it, especially at 10x the cost.
A fun game! I'm not convinced the AI took a particularly sensible path. It would be interesting to see how well different prompts/models succeed at this task.
Very interesting. What I'd really love to see (though perhaps it is too soon to say) is whether these items actually sell. That is, is the output actually something people want to buy?