Does Gary do any AI research these days, or is he full-time Tweet-reactor?

It's got to be exhausting having to write about why each new mind-blowing advance in deep learning isn't actually impressive. The best recent example was Sora. His financial backers would've absolutely creamed themselves if his line of work had produced that, but he was nitpicking this insane development with a magnifying glass.