There's mounting evidence that people who use LLMs for work are less efficient, not more. Even if your work is pushing you to use them, you shouldn't feel like you need to reevaluate your biases because, by all metrics, your discomfort with using them is well-founded.
LLMs are a neat technology, and if they were actually sold based on their merits instead of being sold as a panacea to humanity's ills they wouldn't be so off-putting. LLMs are basically very fancy parrots that can remix what's presented to them in novel ways, and that's really neat, but that's all it is. They lack any context for the things they produce, they're not *intelligent* and they're not bringing us meaningfully closer to actual AI. And that would be all fine and good, except they're being over-promised, over-hyped, and forced into applications that don't suit them. Worse yet, they're being used by corporations as yet another tool to scam consumers with poorly made slop while supplanting an already exploited workforce who create a far better product. It's also abhorrent how they're being promoted less by demonstrating what they're actually capable of and more by creating a cult of personality around them. If you ignore literally all of the context around LLMs they're an impressive technology, but ignoring the context is really harmful. And none of this is even getting into their ecological impact or their impact on low income (largely minority) communities who are forced to live near the data centers.
It's fine for your husband to think it's a really cool technology, but if he fights with you about your own feelings about LLMs then he's ignoring a fuckload of context.
