The Jagged Frontier
5 min read · 4 principles · apply it by lunch
Good morning. How hard a task feels to you predicts almost nothing about how well the machine will do it.
Of the work you handed to a model last week, which outputs did you genuinely check and which did you wave through because the writing was good? Until you can split those two lists, you don’t have a map of the frontier. You have a run of luck.
In 2023 Fabrizio Dell’Acqua of Harvard Business School, Ethan Mollick of Wharton and seven colleagues designed a field experiment, run by Boston Consulting Group on its own people and published as ‘Navigating the Jagged Technological Frontier’. Seven hundred and fifty-eight consultants took part across two separate experiments, randomly assigned inside each to work unaided, with GPT-4, or with GPT-4 plus a briefing on prompting. Inside the model’s competence, the assisted group completed 12.2 per cent more tasks, 25.1 per cent faster, and scored around 40 per cent higher on quality. On a business problem built to sit just outside it — interview notes against a spreadsheet that contradicted them — the model’s users were 19 percentage points less likely to reach the right answer than those working alone. Two halves of one firm, the same model, results running in opposite directions. Capability isn’t a rising tide but a jagged coastline, and two tasks either side of an inlet look identical from the shore. The trap is borrowing somebody else’s map, because the jaggedness takes the shape of the work you personally do.
The Jagged Frontier is Day 99 of 365.
In the Daily, each of its four principles is explained in full, with a one-page template to apply it the same day. A new framework arrives every morning at 7am for a year, across all fifteen courses of the FutureMBA curriculum.
$29 a month · cancel any time · 7-day money-back guarantee