The BaseResearchOrganization Science
Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of Artificial Intelligence on Knowledge Worker Productivity and Quality
In its abstract, Dell'Acqua and colleagues describe a preregistered experiment with Boston Consulting Group in which 758 knowledge workers were randomly given no AI, GPT-4, or GPT-4 plus prompting guidance. On 18 tasks inside the AI's capabilities, AI users did more, faster and better; on one task outside them, they were 19% less likely to be correct.
Why it matters
The same thing, for different readers.
What it means for you
On 18 knowledge tasks AI users did better and faster, but on one complex managerial task they were more often wrong, so knowing when to trust AI may matter.
For lawmakers
A single tool gave mixed results inside one workflow, which bears on training and oversight for workers who use AI; the evidence is from one collaboration and GPT-4 tools.
For researchers
The jagged-frontier framing and preregistered randomized design offer a template for testing where AI helps or hurts knowledge work; the abstract does not separate the three study arms.
In their own words
Every quotation was found, word for word, in the text we saved from the source. That shows the words are there; it does not make them true, or prove the speaker meant them as quoted.
From the source’s own text.
“For each one of a set of 18 realistic knowledge tasks within the frontier of AI capabilities ranging from creative to analytical tasks, subjects using AI outperformed those not using AI, completing 12.2% more tasks and completing them 25.1% more quickly on average while also delivering solutions of significantly improved quality.”
Found word for word in the source
“However, for a complex managerial task selected to be outside the frontier, subjects using AI were 19% less likely to produce correct solutions compared with those without AI, pointing to potential limitations of AI supporting knowledge workers.”
Found word for word in the source
“We introduce and study the concept of a “jagged technology frontier” to describe the uneven impact of artificial intelligence (AI) capabilities, where AI assistance improves performance for some tasks but worsens it for others, even within the same knowledge workflow and with a seemingly similar level of difficulty.”
Found word for word in the source
How the finding was reached
In our words, from the source. Weigh the finding by how it was made.
- What was done
- A preregistered randomized experiment with 758 knowledge workers, run in collaboration with Boston Consulting Group on management-consulting tasks the authors developed. After a performance baseline on similar tasks, participants were randomly assigned to no AI access, GPT-4 access, or GPT-4 access with a prompt-engineering overview. They did 18 realistic tasks, from creative to analytical, inside the AI's capability frontier, and one complex managerial task chosen to lie outside it. Based on the abstract only; the full text was not readable.
- What it found
- On the 18 tasks inside the frontier, participants using AI completed 12.2% more tasks and finished 25.1% more quickly on average, with significantly higher-quality solutions. On the one task chosen to lie outside the frontier, participants using AI were 19% less likely to produce correct solutions than those without AI.
- What it does not show
- Abstract only, so the paper's own limitations, sample details and results by study arm are not reflected. From the design as described: the outside-the-frontier result rests on a single task chosen by the researchers; the tasks were built by the authors with one firm; and the AI was GPT-4. The abstract gives no effect sizes for the arm with prompting guidance and no intervals.
- Status
- Peer reviewed
The record
- Questions
- What happens to work, income and the economy?
- People
- Fabrizio Dell'Acqua (Wrote it, Digital Data Design Institute, Harvard Business School) · Edward McFowland III (Wrote it, Digital Data Design Institute, Harvard Business School) · Ethan Mollick (Wrote it, The Wharton School, University of Pennsylvania) · Hila Lifshitz (Wrote it, Digital Data Design Institute, Harvard Business School; Warwick Business School) · Katherine C. Kellogg (Wrote it, Sloan School of Management, Massachusetts Institute of Technology) · Saran Rajendran (Wrote it, Henderson Institute, Boston Consulting Group) · Lisa Krayer (Wrote it, Henderson Institute, Boston Consulting Group) · François Candelon (Wrote it, Henderson Institute, Boston Consulting Group) · Karim R. Lakhani (Wrote it, Digital Data Design Institute, Harvard Business School)
- Also covered by
- Harvard Business School
- What we read
- We read the abstract.
- Found by
- A research pass: work, income and the economy, gathered October 13, 2026, via
research-work(tier 1: the source’s own channel or an official record) - Cite this
- Organization Science, “Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of Artificial Intelligence on Knowledge Worker Productivity and Quality”, March 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/organization-science-2026-03-jagged-frontier-ai-experiment
Our notesAbstract only. The article page (INFORMS), SSRN and the Harvard Business School PDF returned 403 or a bot-check response, so no full text was read. The saved text is the publisher's Crossref record (bibliographic data, author affiliations and abstract); the abstract is identical to the one on the Harvard Business School page for the 2023 working paper (HBS Working Paper 24-013, September 2023, 58 pages; https://www.hbs.edu/faculty/Pages/item.aspx?num=64700, which lists the Organization Science version as March-April 2026). The date 2026-03 is Crossref's issue month. The abstract does not say that the 758 participants were BCG consultants (the triage does); check the full text. Three authors are at BCG's Henderson Institute and the study was run in collaboration with BCG; the Crossref record states funding from the Harvard Business School Digital Data Design Institute and its Division of Research and Faculty Development. All quotations are from the abstract.