He runs an analytics company and was specifically touting how his models showed a quick Russian victory, even weeks into the invasion, which was more than just the usual Besserwisserism among western europeans towards eastern Europe. If the whole shtick is that the mainstream is lame but my patented crystal ball actually tells the future from first principles and you then tank even harder than the mainstream, you not only embarrass yourself but reveal your own preferences as well.
And the mainstream predictions in the West of Ukrainian resistance weren't as negative as what he advanced, which is precisely why many refused to believe the Kremlin would actually step on this rake of a war. That he wholeheartedly defended a quick Russian victory made him seem steeped more in the Kremlin's narratives than the West's, even as all imperial powers tend to be hubristic.

The luddites smashed looms not because they were inherently anti-technology, but as a means of industrial action in a time where labor held no formal power at all. As you say yourself, the issue with AI is that it doesn't even work as advertised; industrial looms worked fine. 'AI''s biggest problems are the enormous waste, ecological damage, enshittification and labor discipline that it incurs. If Silicon Valley was actually building Data, the debate would be different, but this dimension seems mostly theoretical for now.