There has been extensive debate in the AI safety community about whether AGI will cause human extinction.* This question asks about how likely that is.

(*The debate has been more nuanced than this: Some think that advanced AI that is highly capable in a narrow set of domains—i.e., not generally intelligent enough to qualify as AGI—could cause human extinction. Meanwhile, others believe that AI will only become extinction-level dangerous once it is “superintelligent”—that is, well beyond AGI-level. For simplicity, though, we ask only about AGI in this question.)