For years, we've been asking the wrong question about artificial intelligence. The debate has centered on whether AI will match or exceed human intelligence — as if cognition were a single ladder we're all climbing, with machines gaining on us rung by rung.

But according to researchers challenging this framework, that's not how AI works at all. Enter "jagged intelligence," a concept that's quietly reshaping how technologists and economists think about what these systems can and can't do.

The term describes AI's maddeningly inconsistent performance across different tasks. Large language models like ChatGPT can pass the bar exam, write poetry that moves people to tears, and generate working code in seconds. Then they'll confidently tell you that 9.11 is larger than 9.9, or fail to count how many times the letter R appears in "strawberry."

This isn't a bug that will get patched out in the next update. It's fundamental to how these systems work.

The Jagged Frontier

According to the New York Times, the "jagged intelligence" framework emerged from researchers studying how AI capabilities don't progress uniformly. Instead of a smooth curve of improvement, AI systems exhibit a jagged frontier of abilities — excelling in some areas while remaining surprisingly incompetent in others.

Think of it like a mountain range viewed from above. Human intelligence might look like rolling hills of relatively consistent elevation. AI intelligence looks like the Himalayas next to Death Valley — towering peaks of capability adjacent to deep valleys of failure.

This matters because it fundamentally changes the automation conversation. The old model suggested AI would replace jobs from "bottom to top" — starting with routine tasks and gradually moving up the skill ladder. Jagged intelligence suggests something messier and less predictable.

Winners and Losers in the Jagged Age

The implications for workers are profound and counterintuitive. Some highly-skilled professionals may find AI handles parts of their job better than they do, while struggling with tasks a junior employee could manage. A lawyer might use AI to draft complex contracts but still need a paralegal to organize the filing system.

Creative professionals are already living this reality. AI can generate serviceable marketing copy or stock images in seconds, putting pressure on entry-level creative work. But it still can't art-direct a photoshoot, navigate client politics, or know when to break the rules for effect.

The jobs most at risk aren't necessarily "low-skill" ones — they're jobs where the valuable tasks happen to fall on AI's peaks rather than in its valleys. Data entry might be vulnerable not because it's simple, but because it's the type of pattern-matching AI happens to excel at.

Meanwhile, jobs that seem automatable on paper may prove resilient because they require navigating AI's jagged terrain. A good executive assistant doesn't just schedule meetings — they read between the lines of office dynamics, know when the boss needs an excuse to leave a call early, and understand that "let's circle back" means different things from different people.

Rethinking the Intelligence Comparison

The human-versus-AI framing has always been a bit of a category error, like comparing a calculator to a mathematician. Yes, the calculator is better at arithmetic, but that doesn't mean it's "smarter" — it's a different kind of tool entirely.

Jagged intelligence formalizes this intuition. It suggests we should stop asking "is AI as smart as humans?" and start mapping the specific peaks and valleys of what it can do. The question isn't whether AI will replace human workers, but which human tasks fall on which side of the jagged frontier.

This reframing also helps explain why AI predictions have been so consistently wrong. Tech evangelists keep declaring that artificial general intelligence is just around the corner, while skeptics point to AI's obvious limitations as proof it's all hype. Both camps are using the wrong measuring stick.

The reality is more nuanced. AI will continue advancing in jagged, unpredictable ways — suddenly conquering tasks we thought were uniquely human while remaining stumped by things a child could do.

The Policy Challenge

For policymakers and business leaders, jagged intelligence presents a thorny challenge. You can't just identify "at-risk occupations" and retrain those workers. The disruption will be more granular and harder to predict.

A radiologist might find AI handles routine scans better than they do, but still be essential for complex cases and patient communication. A software engineer might use AI to write boilerplate code while spending more time on system architecture and debugging AI-generated mistakes.

The winners in this transition will be workers and organizations that learn to navigate the jagged frontier — understanding what to delegate to AI and what requires human judgment. The losers will be those who either resist AI entirely or blindly trust it in its valley zones.

As the New York Times reporting suggests, this framework could help us move past the binary "AI will take all our jobs" versus "AI will create new opportunities" debate. The truth is both and neither — it will reshape work in ways that don't fit our existing categories.

The sooner we accept that AI intelligence is jagged rather than linear, the better we can prepare for a future where the most valuable skill might be knowing which peaks to climb and which valleys to avoid.