Because Autocorrect doesn’t make decisions or analyses?

Unlike coding, LLMs struggle with video games. This feels surprising given their success in coding, as well as in games like chess and Go.

Can’t really take the article seriously with statements like this.

Last time I checked, LLMs are comically bad at chess. They can do some openings but then very quickly degenerate into illegal moves and hallucinations about which pieces are where. There was a popular article last year wherein Chess for the Atari 2600 beat ChatGPT in a head to head match, and I don’t think the situation has improved much since then.

Why is a screwdriver really bag at hitting a nail?

Just at video games? Hmm.