Tag
This thread discusses the concept of 'Jagged Intelligence' in AI, framing it as a consequence of AI learning being an ill-posed inverse problem, and argues that external stabilizers like scaffolding and verification are essential.
The tweet comments on the concept of jagged intelligence, noting that code is an exception to the pattern where fable is not significantly better.
The article discusses the concept of 'jagged intelligence' from Andrej Karpathy, highlighting the uneven distribution of AI capabilities across domains and arguing that the true value lies in the 'harness'—the domain-specific engineering and tooling built around generalist models. It asserts that small teams with deep domain expertise can achieve significant asymmetric advantages, particularly in cybersecurity.