Yeah but it was only like 2 years ago that artists were arguing this on the basis that AI-gen images would consistently mangle hands

Now we're at a point where that never happens, and where lipsync is almost a completely solved problem

If the issue here is simply that the quality is bad, one has to contend with the fact that it is undoubtedly exponentially improving and there's no reason we should expect that improvement to stop

I also don't have any interest in consuming AI generated art, but the same criticisms were levied at computer graphics and if we're comparing to CGI we'd be at the late 1970s in terms of nascency

I've made ai generated art using family photos as the starting point, and it was wonderful. :)

I'm sure most engineering is LLM-assisted already and nothing is wrong with it. It's just the one-shot vibe-coded low quality slop that spoils sentiment of this tools. Also many people are interested in what agents can build unsupervised as a test of "superintelligence".