What is the technological singularity? AI and self-improvement
A technological shift that changes the basis of prediction
The technological singularity is the idea that intelligence beyond human levels could accelerate technological and social change so much that extrapolating from the past becomes unreliable. Vernor Vinge discussed such a transition in a 1993 essay.
AGI and ASI describe capabilities; singularity describes change
AGI and ASI mainly describe capabilities. The singularity concerns the speed and scale of changes such intelligence could cause. Calling a system AGI does not, by itself, mean society must immediately transform.
What if AI helps build the next AI?
One possible route to acceleration is a cycle in which AI helps research, the findings improve the next AI, and that AI speeds up further research. Generating ideas and completing experiments or manufacturing are different steps. The term self-improvement alone does not demonstrate unlimited acceleration.
What to check in the news
- What parts of future AI research and development are assigned to AI?
- Do improvements feed back into the next research cycle?
- Does a forecast explain the mechanism, rather than only name a year?
An imagined future starts here
Imagine AI helping design better computers, which then support further research. If tools lead to better tools at shorter intervals, schools and companies might shift from repeatedly using established methods to continually trying new ones. The Future Life Atlas imagines how such changes could reach everyday life.
Future Life AtlasRead more on this topic
Could AI research results improve the next AI?
AI helps mathematical research, and its results could help improve AI. We explore how this cycle might develop.
Could independent evaluation make new AI easier to adopt?
Shared evaluation procedures could help businesses and laboratories decide when to adopt new AI.