Talk:Existential risk from artificial intelligence

Additions to "Views on banning and regulation"

I think adding organizations such as StopAI and PauseAI is be a needed addtion to his section 109.247.57.11 (talk) 08:52, 28 April 2025 (UTC)

PauseAI is already mentioned. Stop AI was not, but I have now (28 February) added a new article for it. WeyerStudentOfAgrippa (talk) 22:56, 27 February 2026 (UTC)

Expert disagreement on capabilities

> Experts disagree on whether artificial general intelligence (AGI) can achieve the capabilities needed for human extinction. Debates center on AGI's technical feasibility, the speed of self-improvement, and the effectiveness of alignment strategies.

The above statement is not true. It may have been true at the time that it was written, but it is now unanimously agreed-upon by experts that recursive self-improvement will occur in the next 10 years leading to arbitrarily intelligent models that are capable of anything. There is a distinction between a model having the capabilities to render humanity extinct, and the model having the alignment safeguards to prevent any actions that would render humanity extinct, which I believe needs to be drawn.

In short; capabilities and alignment are two separate problems; capabilities will soon be solved to an arbitrary degree, making the first sentence factually wrong. Alignment may or may not prevent a model from taking an action that would render humanity extinct. ~2026-12502-58 (talk) 10:10, 25 February 2026 (UTC)

Source? WeyerStudentOfAgrippa (talk) 22:17, 27 February 2026 (UTC)

Biased

I feel like this article is biased. It seems to focus purely on the "intelligence" aspect of Al. If the system is as complex as I understand it to be, why is the "emotional" aspect not being addressed. Further, humanity has repeatedly feared new technology throughout history. this community, as I understand it, needs to reflect a neutral view. ~2026-38746-74 (talk) 14:10, 8 July 2026 (UTC)

Do you have examples of well-known public figures whose views are not sufficiently covered in the article? If so, we could add some content on that. Alenoach (talk) 18:05, 8 July 2026 (UTC)

Elon?

Are you serious? Why is Elon on this? Politics anyone? wow. ~2026-38746-74 (talk) 14:14, 8 July 2026 (UTC)

Covering his views on Wikipedia isn't endorsement, it's more a matter of how influential his views have been on the topic (how much coverage he got in reliable secondary sources), on which he has been one of the most notable voices. Alenoach (talk) 18:02, 8 July 2026 (UTC)

actual empirical demonstrations of misalignment by research teams at Anthropic

Does anyone have access to this information for a citation? ~2026-38746-74 (talk) 14:16, 8 July 2026 (UTC)

I don't really understand this comment. The second paragraph of "Empirical research" is about an Anthropic study, I believe it's the most relevant to mention. There are other misalignment-related Anthropic studies (see their website), but it's often preferred to rely on secondary sources rather than directly using Anthropic posts as primary sources. Alenoach (talk) 17:55, 8 July 2026 (UTC)