Managing Intelligence: Skilled Experts and AI in Markets for Complex Products
Open accessThe DOI matches its Crossref record; 7 of the 7 references it cites were found in a registry or an index.
Abstract
In numerous high stakes markets skilled experts play a key role in facilitating consumer choice of complex products.New artificial intelligence (AI) technologies are increasingly being used to augment expert decisions.We study the role of technology and expertise in the market for health insurance, where consumer choices are widely known to be sub-optimal.Our analysis leverages the large-scale implementation of an AI-based decision support tool in a private Medicare exchange where consumers are randomized to skilled agents over time.We find that, prior to AIbased technology, skilled experts in this market exhibit the same type of inconsistent behavior found in previous studies of individual choices, costing consumers $1260 on average.The addition of AI-based decision support improves outcomes by $278 on average and substantially reduces heterogeneity in broker performance.Experts efficiently synthesize private information, incorporating AI-based recommendations along dimensions that are well suited to AI (e.g. total expected patient costs), but overruling AI-based recommendations along dimensions for which humans are better suited (e.g.specifics of doctor networks).As a result, switching plans, an expost measure of plan satisfaction, is meaningfully lower for agents making AI-based recommendations.While AI is a complement to skill on average, we find that it is a substitute across the skill distribution; lower quality agents provide better recommendations with AI than the top agents did without it.Overall productivity rises, with the introduction of decision support associated with a 21% reduction in call time for enrollment.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Automation and New Tasks: How Technology Displaces and Reinstates Labor
Acemoglu, D., & Restrepo, P. (2019) · Journal of Economic Perspectives · doi:10.1257/jep.33.2.3 · Cited by 2,612
Human Decisions and Machine Predictions
Kleinberg, J., Lakkaraju, H., Leskovec, J., Ludwig, J., & Mullainathan, S. (2018) · The Quarterly Journal of Economics · doi:10.1093/qje/qjx032 · Cited by 1,040
Testing for Asymmetric Information in Insurance Markets
Chiappori, P., & Salanie, B. (2000) · Journal of Political Economy · doi:10.1086/262111 · Cited by 1,013
Adverse Selection and Inertia in Health Insurance Markets: When Nudging Hurts
Handel, B. R. (2013) · American Economic Review · doi:10.1257/aer.103.7.2643 · Cited by 707 · cited 8×
“…parameters to customers based on responses to survey questions about attitudes towards risk, but in this particular application, such survey questions were not permitted, so all customers were assigned a risk aversion parameter of = 4.0 ⇤ 10 4, which is similar to estimates from Handel (2013).”
InfluentialSelf-citationHuman Decisions and Machine Predictions
Kleinberg, J., Lakkaraju, H., Leskovec, J., Ludwig, J., & Mullainathan, S. (2017) · doi:10.3386/w23180 · Cited by 545
Health Insurance for “Humans”: Information Frictions, Plan Choice, and Consumer Welfare
Handel, B. R., & Kolstad, J. T. (2015) · American Economic Review · doi:10.1257/aer.20131126 · Cited by 427
Self-citationThe Allocation of Decision Authority to Human and Artificial Intelligence
Athey, S., Bryan, K., & Gans, J. (2020) · doi:10.3386/w26673 · Cited by 14
Cited by
Algorithmic Risk Assessment in the Hands of Humans
Stevenson, M., & Doleac, J. L. (2019) · doi:10.2139/ssrn.3489440 · Cited by 60
OK computer: Worker perceptions of algorithmic recruitment
Fumagalli, E., Rezaei, S., & Salomons, A. (2022) · Research Policy · doi:10.1016/j.respol.2021.104420 · Cited by 46
When Less is More: Improving Choices in Health Insurance Markets
Abaluck, J., & Gruber, J. (2023) · The Review of Economic Studies · doi:10.1093/restud/rdac050 · Cited by 36
Algorithmic risk assessment in the hands of humans
Stevenson, M. T., & Doleac, J. L. (2024) · American Economic Journal: Economic Policy · doi:10.1257/pol.20220620 · Cited by 21
Voice AI in Firms: A Natural Field Experiment on Automated Job Interviews
Jabarian, B., & Henkel, L. (2025) · doi:10.2139/ssrn.5395709 · Cited by 9
Metadata
- Authors
- Gruber, J., Handel, B., Kina, S., & Kolstad, J.
- Venue
- National Bureau of Economic Research
- Publisher
- National Bureau of Economic Research
- Published
- 2020-04
- OpenAlex
- W2943594921
- Cited by
- 35
- References
- 7
- Type
- report
- Language
- en
- Open access
- green
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.