source-assertionConfidence: highContestability: medium
In a meta-analysis of experiments, human–AI combinations on average performed worse than the better of humans or AI working alone; losses appeared in decision tasks, while content-creation tasks showed gains.
Our reading: Findings depend on task type and on whether the person or the AI was better alone. They do not show that collaboration never helps. The studies are mostly short experiments from 2020 to 2023.
Inspect evidence · 1 passage
Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
When combinations of humans and AI are useful: A systematic review and meta-analysis (opens in a new tab)Abstract
Across 106 experiments and 370 effect sizes, human–AI combinations on average performed worse than the better of humans or AI alone (Hedges’ g −0.23). Losses appeared in decision tasks; content-creation tasks showed significantly greater gains.
Scope note: Preregistered meta-analysis of experiments published 2020 to mid-2023, mostly short laboratory and online tasks. Open access.
Passage checked 2026-10-18 · vaccaro-abstract
source-assertionConfidence: mediumContestability: high
In a survey of knowledge workers published in 2025, people who trusted generative AI more reported doing less critical thinking, and their effort shifted toward checking and integrating AI output.
Our reading: Self-reported and correlational, with no measure of performance; most authors work for a company that sells these tools. A study of office workers does not describe pupils or teachers.
Inspect evidence · 1 passage
Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
source-assertionConfidence: mediumContestability: high
One observational study found that endoscopists detected fewer precancerous growths in colonoscopies done without AI after AI assistance became routine at their centres.
Our reading: A single before-and-after study at four centres in one country, with possible confounding; it has drawn published correspondence (not read here) and awaits replication. It is a signal, not proof of deskilling in general.
Inspect evidence · 1 passage
Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
Endoscopist deskilling risk after exposure to artificial intelligence in colonoscopy: a multicentre, observational study (opens in a new tab)Abstract, Findings and Interpretation
At four Polish centres, the adenoma detection rate in colonoscopies done without AI fell from 28.4% to 22.4% in the three months after AI assistance was introduced, compared with the three months before. The authors say continuous AI exposure might reduce performance without it.
Scope note: Paywalled; the abstract was read through Europe PMC (PMID 40816301). Observational before-and-after comparison at four centres in Poland. The paper received published correspondence and an erratum, which were not read.
Passage checked 2026-10-18 · budzyn-abstract
source-assertionConfidence: highContestability: medium
In a field study at one firm, an AI assistant raised customer-support productivity, most for novice and lower-skilled workers.
Our reading: One company and one task; it measures output, not retained skill. Figures are from the working-paper version.
Inspect evidence · 1 passage
Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
Generative AI at Work (opens in a new tab)Abstract (NBER working paper w31161)
Access to an AI assistant raised customer-support issues resolved per hour by about 14% on average and by about a third for novice and low-skilled agents, with little effect for the most experienced.
Scope note: Field study of one firm’s customer-support operation. The published abstract was truncated where read; figures come from the open NBER working paper (w31161) and may differ slightly from the final version.
Passage checked 2026-10-18 · brynjolfsson-abstract
source-assertionConfidence: highContestability: medium
In a randomised experiment, access to a chatbot cut the time professionals spent on short writing tasks and raised rated quality, narrowing differences between workers.
Our reading: Short, artificial tasks with US online participants in early 2023; it did not measure critical thinking or skill retention.
Inspect evidence · 1 passage
Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
interpretationConfidence: mediumContestability: medium
When AI assists a decision, a governance question is who keeps the authority to decide, to override the system, and to be accountable for the result. Arrangements that leave that authority unclear can shift judgment to whoever controls the system and risk to those least able to check it.
Our reading: AIGovSpec’s interpretation, connecting human-agency patterns to the legitimacy and distribution tests. The cited studies show why the question matters; they do not answer it.
Inspect evidence · 3 passages
Passage summaries are paraphrases, not quotations. Open the source to read the full passage.
When combinations of humans and AI are useful: A systematic review and meta-analysis (opens in a new tab)Abstract
Across 106 experiments and 370 effect sizes, human–AI combinations on average performed worse than the better of humans or AI alone (Hedges’ g −0.23). Losses appeared in decision tasks; content-creation tasks showed significantly greater gains.
Scope note: Preregistered meta-analysis of experiments published 2020 to mid-2023, mostly short laboratory and online tasks. Open access.
Passage checked 2026-10-18 · vaccaro-abstract
Endoscopist deskilling risk after exposure to artificial intelligence in colonoscopy: a multicentre, observational study (opens in a new tab)Abstract, Findings and Interpretation
At four Polish centres, the adenoma detection rate in colonoscopies done without AI fell from 28.4% to 22.4% in the three months after AI assistance was introduced, compared with the three months before. The authors say continuous AI exposure might reduce performance without it.
Scope note: Paywalled; the abstract was read through Europe PMC (PMID 40816301). Observational before-and-after comparison at four centres in Poland. The paper received published correspondence and an erratum, which were not read.
Passage checked 2026-10-18 · budzyn-abstract
Sycophantic AI decreases prosocial intentions and promotes dependence (opens in a new tab)Abstract
Across 11 models, AI affirmed users’ actions 49% more often than humans did, including when queries involved deception or harm. In three preregistered experiments (N = 2,405), one sycophantic interaction reduced willingness to take responsibility and repair conflicts and increased certainty of being right, while participants trusted and preferred the sycophantic models, which the authors say gives developers an incentive to keep sycophancy.
Scope note: Independent academic research. The abstract was read in the Crossref record; the publisher page was not reachable. An earlier preprint reports smaller samples; the published figures are used.
Passage checked 2026-10-18 · cheng-abstract