[vsl]

But people treat them as if they could. You can already see them using LLM output as "proof" in discussions...

[voidspark]

They can simulate thinking or reasoning well enough, better than some humans.