Experimental evidence on the productivity effects of generative AI

Citation

Publisher

Science (Noy & Zhang, MIT)

Title

Experimental evidence on the productivity effects of generative AI

Date

July 2023

Provenance

Independent research

Method

Randomised controlled trial, 453 college-educated professionals, blind grading.

What it does not show

Short, self-contained tasks. Says nothing about campaign-level outcomes, brand consistency, or the review time a regulated market adds.

Full reference

Science (Noy & Zhang, MIT). Experimental evidence on the productivity effects of generative AI. July 2023. https://www.science.org/doi/10.1126/science.adh2586

Where this source appears

Cited on 2 field guides. The claim drawn from it on each one is reproduced below, so you can check the reading against the source itself.

• Marketing

  A randomised trial of 453 writers on mid-level professional writing tasks found time on task fell 40% and output quality rose 18%.

  Read the marketing guide

• Content creation

  A randomised trial of professional writing tasks found a 40% reduction in time and an 18% rise in graded quality.

  Read the content creation guide

We take no payment from any organisation named on this page. If a detail here is wrong or out of date, tell us and we will correct it with the date of the change.