Definition
The uneven and invisible line between tasks AI does well and tasks it does badly, which users cannot reliably locate from the inside.
What is at stake
It is the reason a pilot that works is not evidence the next task will, and it is the objection the first sceptic in the room reaches for.
Where the word came from
- The jagged frontierthe jagged frontier study, 2023
- What explains the jaggednessartificial general cleverness, Terence Tao, December 2025. Tao is a mathematician at UCLA and a Fields Medallist, writing about AI systems as someone who uses them on hard problems. These systems solve broad classes of complex problems by ad hoc means, stochastic, uninterpretable or lifted from training data, without the understanding or reliability that intelligence implies. In people, cleverness and intelligence go together. In these systems they come apart, which is why the frontier is jagged, and which the 2023 study establishes without accounting for
How episcope came to it
The jagged frontier study: Dell'Acqua, McFowland, Mollick and six colleagues, with BCG, first circulated in 2023 as a Harvard Business School working paper, then peer reviewed and published in Organization Science in 2026. 758 consultants, 18 tasks inside the frontier and one placed outside it. Inside, work rated over 40 percent higher in quality. Outside, consultants using AI were 19 percentage points less likely to reach the correct answer. The explanation came two years later from outside the study, in Tao's post.
What to watch for
The frontier moves with every model release. The study's finding about the line holds. Its map of where the line ran in 2023 is dated, which is the point of the word invisible. Tao's contribution is a post on Mathstodon, 15 December 2025, and not a paper, so it is cited as his post with the date.
Papers and articles behind it
Dell'Acqua, F., McFowland III, E., Mollick, E., Lifshitz-Assaf, H., Kellogg, K., Rajendran, S., Krayer, L., Candelon, F., & Lakhani, K. R. (2026). Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality. Organization Science, 37(2).