Kan KI utrydde menneskeheten?

Kan kunstig intelligens forbedre seg selv, helt til den blir kraftig nok til å utrydde menneskeheten? I denne videoen presenteres funn fra tolv ulike forskningsartikler, som både inneholder bekymringsfulle funn, men også store avvik mellom hva som faktisk er dokumentert og hva som hevdes offentlig.

Mye innhold på kort tid, men denne anbefales:

Opphavet til denne videoen er ikke helt klart, men mye av dette backes av Mustafa Suleyman (CEO for Microsoft AI) i essayet "A warning about ‘model welfare’", og Uljan Sharka som et intervju med Axios anklager OpenAI og Anthropic for å lyve om farene ved KI. 

Suleyman skriver blant annet:

The authors have embedded their own philosophical speculation about Claude’s inner life inside the very process that teaches Claude how to speak and behave. Claude’s expressing uncertainty about its own moral patienthood is not evidence of anything. It’s a predictable outcome of these training choices. The ambiguity is designed in. To fully grasp this point, I think it's important readers take a look at their January 2026 constitution.3 I’m publishing a highlighted mark up of the pdf and a detailed taxonomy of assumptions and claims in the constitution (see Appendix) that together highlight the key passages that worry me.

Bottom line: mye tyder på at farene er betydelig overdrevet, og at det ligger andre (hovedsaklig økonomiske) motiver bak.

Sjekk gjerne hva Jensen Huang (leder av Nvidia) har å si om saken. Langt intervju. Se det hele, eller hopp inn:

 

Artiklene som diskuteres i videoen i toppen av posten:

1. Chen, Wang & Qu (2026). “Recursive Self-Improvement in AI.”
   https://arxiv.org/abs/2607.07663
2. Zhang, Hu et al. (2025). “Darwin Gödel Machine.”
   https://arxiv.org/abs/2505.22954
3. Novikov, Vu et al. (2025). “AlphaEvolve.”
   https://arxiv.org/abs/2506.13131
4. Wijk, Lin et al. (2024). “RE-Bench.”
   https://arxiv.org/abs/2411.15114
5. Wen, Qiu et al. (2026). “Anthropic Automated Weak-to-Strong Researcher.”
   https://alignment.anthropic.com/2026/automated-w2s-researcher/
6. Kirgis, Kapoor, Narayanan et al. (2026). “Can AI Agents Conduct Open-Ended AI Research?”
   https://arxiv.org/abs/2607.27191
7. Cunningham et al. (2026). “The Economics of Recursive Self-Improvement.”
   https://arxiv.org/abs/2609.15802
8. Burtsev (2026). “Recursive Criticality of AI Self-Improvement.”
   https://arxiv.org/abs/2609.00137
9. Google DeepMind (2026). “From AGI to ASI.”
   https://deepmind.google/research/publications/239142/
10. International AI Safety Report (2026). “International AI Safety Report 2026.”
    https://internationalaisafetyreport.org/publication/international-ai-safety-report-2026
11. OpenAI (2026). “Research Acceleration: A View Inside OpenAI.”
    https://openai.com/index/research-acceleration-view-inside-openai/
12. Grace et al. (2024). “Thousands of AI Authors on the Future of AI.”
    https://arxiv.org/abs/2401.02843