But what need does anyone have for a "coherent and relevant text" that isn't true?

It is correctness that we should be prioritizing.

You don't even discuss the idea. How "fine tuning on specific tasks" or "ensembles of models" will somehow allow these programs to somehow distinguish truth from falsehood in these huge corpora is absolutely not clear.

I expected to be pro-AI. But now I see how it's shaking out, I am completely against it.

It's not just the lying and confabulation - it's that it's yet another heist by big companies. They have taken all this information that we individual humans put into the world, and used it to build proprietary models owned by massive companies who hope to exploit it for maximum profits.

Of course, this has been the history of the last two generations. An unparalleled amount of new wealth has been generated, and all of it has gone to the richest 1%, while the lives of almost everyone else has become more and more difficult and precarious.

When I was young, you could graduate from high school, get a job in a factory, support a stay-at-home partner and children, send those kids to university, and then retire in modest comfort.

Now two parents working full-time will not be able to achieve that.

And the reason this is true is because all the excess value got vacuumed up by a tiny number of very rich people, mostly clinical sociopaths.

This is yet another step in the wrong direction. The 1% simply refuse to negotiate or relent in their steady destruction of our economy, our climate and our ecosystem. We the people need to take matters into our own hands.