The Transformative Potential of AGI — and When It Might Arrive | Shane Legg and Chris Anderson | TED
AI Summary
- Early Interest in AI
- First computer at age 10, pre-internet era
- Began programming and creating virtual worlds
- Spark of creativity led to interest in AI
- Educational Challenges
- Dyslexic, nearly held back in school
- High IQ discovered during assessment
- Early skepticism of authority and interest in intelligence
- Coined “artificial-general-intelligence” (AGI)
- Term suggested to Ben Goertzel in 2001 for a book title
- Term popularized online, though Mike Garrod first used it in 1997
- Prediction of AGI
- Influenced by Ray Kurzweil’s book
- Predicted exponential growth in computation and data
- Predicted AGI by 2028 with a 50% chance
- Definition of AGI
- Initially, a vague concept of general systems
- Now defined as systems capable of all cognitive tasks humans can do
- Founding of DeepMind
- Met Demis Hassabis at Gatsby Unit in London
- Founded DeepMind with the goal of creating AGI
- Aware of both positive and negative potential outcomes
- DeepMind’s Progress
- Milestones: Atari games, AlphaGo, language models
- Surprised by the capabilities of text-based models
- Concerns and Hopes for AGI
- Believes powerful AI is coming, but timing uncertain
- Potential for rapid scientific advancements
- Acknowledges risks and need for ethical considerations
- Regulation and Understanding of AI
- Advocates for regulation of powerful technologies
- Emphasizes the need for more scientific understanding of AI
- Hopes for ethical and safe AI to usher in a new golden age for humanity