As you say, responsible ML is an interpretable scalpel, not a black box hammer. These ocean boiling LLMs purporting to be a tool with universal generalism are such a categorical disaster–for ecology, for the health of the field, for public understanding.

It’s hard to go to work and have these conversations over and over again, to be honest it’s depressing, wearing me down. I suspect I’ll say no one too many times to “putting AI in our product” (by which they mean destroying simple well designed navigation by putting everything behind a chatbot search etc) and be let go.