AI ‘could be last technology humanity ever builds’, expert warns in ‘doom timeline’
Expert issues dire AI warning and ‘doom timeline’
- A research project, AI 2027, initially predicted that artificial intelligence could develop "superintelligence" and autonomous coding capabilities by 2027, potentially leading to human domination.
- AI Futures, the group behind the study, has since revised this timeline, stating that AI's development of autonomous coding is now expected in the 2030s, with superintelligence following later.
- Project leader Daniel Kokotajlo explained the revision, noting that AI development appears to be progressing slower than initially anticipated.
- While some experts, like Gary Marcus, dismissed the original timeline as "pure science fiction", others acknowledge the importance of discussing AI's potential risks.
- Dr Fazl Barez from the University of Oxford warned that AI could be "the last technology humanity ever builds" if safety measures and alignment are not adequately addressed, stressing that AI should serve humanity.