Summary of what Chomsky says, in my own words:

I’m doing science dammit; those programs are engineering. Apples and bananas. I’m scientifically interested on how humans develop language, and LLMs are designed in a way that don’t tell you shit about that.

Tom Jones be like “hey tchonski your refuted lol lmao haha”. He ignores how science works; no, I don’t need to sample every bloody language to draw hypotheses on how Language works.

Side note for a whole century we’ve been bombarded with corporate “fuck science!” campaigns. Capitalism seeks profit, not human welfare.

Chatbots cannot in principle match the linguistic competence of humans, because their basic design prevents them from reaching the minimal condition for that: to distinguish between possible and impossible languages. However, it’s possible that future engineering projects might match or even surpass human performance in this aspect. Like automatic calculators for example. Or other animals, like, our navigation abilities suck compared with ants.

No, this shit does not have moral faculties. Yes, AI engineering can pose severe threats. We should try it but the cat is out the bag, I’m sceptical about the possibility of doing so, malicious actors can probably find ways to evade safeguards.

Side note Chomsky rambles a bit too much even for my own standards. What he writes is often worth reading, regardless of agreeing or disagreeing with him; but come on, get to the point dammit! Also note that he doesn’t directly address the interviewer’s questions, only indirectly.