Topics / Artificial intelligence
Does advanced AI pose a serious risk of human extinction or loss of control?
People disagree about whether advanced AI could cause human extinction or a permanent loss of human control, how likely that is, and how much priority it should get compared with other harms and benefits of AI. The disagreement mixes contested evidence about future capabilities with differing values about risk, precaution and present needs.
Positions
- Yes, it should be a global priority: AI systems that come to exceed humans across most domains could pursue goals we did not intend, and we do not yet know how to reliably control them. Reducing the risk of extinction or permanent loss of control should be a global priority alongside pandemics and nuclear war.
- Take it seriously, but in proportion to the uncertainty: Nobody knows how likely catastrophic loss of control is. The sensible response is to fund safety research, monitor the most capable systems and build the capacity to act, scaling the response to the evidence as it emerges rather than either dismissing the risk or treating it as the overriding priority.
- Present harms matter more: Bias, misinformation, surveillance and the concentration of power in a few companies are harming real people now, and they deserve priority. A focus on speculative extinction scenarios risks distracting from these harms and can serve the interests of the largest AI firms.
- The risk is overblown: Today's AI systems are tools without goals of their own, which humans build, deploy and can limit or switch off. Scenarios of extinction or loss of control rest on speculation rather than evidence and should not drive policy.