jagged intelligence
An AI that is excellent at some tasks but badly flawed at others, lacking consistent general competence.
As defined in Demis Hassabis #2
Other definitions
AI systems that exhibit uneven capability profiles—performing exceptionally well on certain formulations of a problem but failing at elementary things when posed a slightly different way. A term coined by Hassabis to describe inconsistency in current AI systems.
Demis Hassabis: Why AGI is Bigger than the Industrial Revolution & Where Are The Bottlenecks in AIThe idea that AI is superhuman in some tasks and very weak in others, without a smooth overall capability profile.
Sam Altman - How to Make an Abundant Future - [Invest Like the Best, EP.484]
Where it comes up · 3
- The future of intelligence | Demis Hassabis (Co-founder and CEO of DeepMind)2 claims
- Demis Hassabis #21 claim
- 343 | Tom Griffiths on The Laws of Thought1 claim
Most important claims · 4 of 4Search all claims for “jagged intelligence” →
- Current AI systems exhibit jagged intelligence: they can achieve PhD-level performance on some tasks like International Math Olympiad problems yet fail on trivial high-school-level logic problems, and this inconsistency is a major gap that must be closed before AGI.The future of intelligence | Demis Hassabis (Co-founder and CEO of DeepMind)
- Current AI systems differ from human minds in two big ways: inductive bias (ability to learn from small data) and generalizability—an AI can be very good at one problem yet fail spectacularly on an adjacent one, a pattern called jagged intelligence; cognitive science is a good tool for understanding where those jagged boundaries lie and improving the systems.343 | Tom Griffiths on The Laws of Thought
- There is roughly a 50% chance of achieving AGI within the next five years (by 2030), where AGI is defined by a high bar of matching the brain's general cognitive functions consistently across all domains—not today's jagged, inconsistent intelligence—plus possessing genuine invention and creativity capabilities that current systems lack.Demis Hassabis #2
- Current AI systems exhibit 'jagged intelligence'—performing at PhD-level on some tasks (e.g., winning International Math Olympiad medals) while failing at high-school level on others (basic logic, chess)—indicating inconsistency that AGI systems should eventually resolve.The future of intelligence | Demis Hassabis (Co-founder and CEO of DeepMind)