He’s been consistently voicing realistic statements with regards to AGI (he doesn’t believe it’s a thing), AI safety and doomerism, limitations of LLMs, etc.
Lost respect for his opinions massively after seeing how he lied about the topic.
And that’s true now as ever. I also heard him say that training multimodal models on text+image/video would mitigate the grounding issue, and that’s proven to be true too.
So I’m not sure exactly what your objection is.
His research at meta is in analytical methods of AI so I'm not surprised. LLMs were seen as anathema to him as he could not comprehend the capabilities that would come from simply scaling up a suitable architecture with compute and data.
It's quite remarkable how confidently incorrect he was proven to be. Now he's going for round two on his takes with AI safety. Clearly he didn't learn his lesson.