AI could make scientists do more work less well, not less work better, study argues
A new study models how AI could paradoxically degrade research quality: saved time gets redirected into quantity, not depth.
Why it matters
As AI adoption accelerates in knowledge work, this research challenges the assumption that time-savings improve outcomes. For practitioners deploying AI in research-intensive orgs, it flags a potential productivity trap: more papers, weaker results.
The key facts
8 to knowTheoretical study argues AI time-savings can redirect effort toward quantity over quality
In 2 of 3 modeled scenarios, individual publication quality drops despite time savings
Researchers' freed hours get funneled into starting new projects rather than improving existing work
Study assumes language models work perfectly—quality degradation is structural, not capability-based
Theoretical study models three scenarios; in two, individual publication quality drops despite AI time savings
Mechanism: saved time becomes economically valuable, redirecting effort to new projects rather than improving existing research
Applies to any knowledge work with similar incentive structures (consulting, engineering, analysis)
Suggests productivity gains from AI are not destiny—organizational policy and metrics drive outcomes
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: Even if language models worked perfectly, they could make research worse, not better. A new theoretical study argues that because AI saves time, researchers' remaining hours become more valuable and get funneled into starting new projects instead of improving existing ones. In two out of three…