pith. sign in

arxiv: 2606.21880 · v1 · pith:J5EPZMEOnew · submitted 2026-06-20 · 💰 econ.GN · cs.CY· q-fin.EC

Human Capital, AI, and Labor Commoditization

Auyon Siddiq , Niuniu Zhang This is my paper

Pith reviewed 2026-06-26 11:13 UTC · model grok-4.3

classification 💰 econ.GN cs.CYq-fin.EC
keywords human capitalAI exposurelabor commoditizationonline labor marketsdifference-in-differencesChatGPTtext embeddingslabor demand
0
0 comments X

The pith

In AI-exposed job categories, human capital becomes less important for predicting labor demand while price becomes more important.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper asks whether generative AI changes how labor markets value human capital. It represents Upwork worker profiles with text embeddings, measures how much human capital information and price each predict labor demand, and applies a difference-in-differences design timed to the ChatGPT release. In more AI-exposed categories the weight on human capital falls and the weight on price rises. The same pattern appears in lower demand premiums for high-human-capital workers and in demand shifting toward lower-priced workers. These patterns indicate that AI produces a commoditization effect on labor with direct consequences for worker skill investment and platform design.

Core claim

Using high-dimensional text embeddings of worker profiles from Upwork and a difference-in-differences design around the release of ChatGPT, the analysis shows that in more AI-exposed job categories the importance of human capital information in predicting labor demand declines while the importance of price rises. Supporting evidence includes a reduced demand premium for workers with strong human capital and a reallocation of demand toward lower-priced workers.

What carries the argument

Text embeddings of worker profiles combined with difference-in-differences estimation around the ChatGPT release to track changes in the predictive importance of human capital versus price.

If this is right

  • Demand premiums for strong human capital fall in AI-exposed categories.
  • Demand reallocates toward lower-priced workers in those categories.
  • Workers face changed incentives to invest in human capital.
  • Online labor market design must account for reduced differentiation on skill attributes.
  • Labor welfare outcomes shift as price competition intensifies.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Education and training priorities may shift away from skills that AI can substitute.
  • Platforms could introduce new quality signals to counteract pure price competition.
  • The pattern may extend to other generative AI tools beyond ChatGPT.
  • Wage compression could appear in sectors with high AI exposure.

Load-bearing premise

Trends in labor demand would have remained parallel between AI-exposed and non-exposed job categories in the absence of ChatGPT.

What would settle it

If pre-ChatGPT trends in the measured importance of human capital already diverged between exposed and non-exposed categories, the causal claim would not hold.

Figures

Figures reproduced from arXiv: 2606.21880 by Auyon Siddiq, Niuniu Zhang.

Figure 1
Figure 1. Figure 1: Sample Construction. Searchable worker profiles (March 2026) 201,857 workers ≥ 1 contract opened in 2022 50,598 workers ≥ 1 job subcategory recorded 49,610 workers Balanced worker-quarter panel 2021Q1–2026Q1 1,041,810 worker-quarters Excluded: No 2022 contracts 151,259 workers Excluded: No subcategory recorded 988 workers We use this sample to construct a balanced worker-quarter panel from 2021Q1 through 2… view at source ↗
Figure 2
Figure 2. Figure 2: Distribution of Contracts over Worker-Quarters [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Worker Profile Information Self-Presentation Title Description Skill Tags Location Credentials Education Employment Portfolio Reputation Contracts Feedback Ratings Badges Price Hourly Rate Worker Profile Blocks Human Capital Signals Nearly all workers have profile titles and descriptions, and all workers have skill tags and posted hourly rates. Credentials are also common: 47,976 workers report education (… view at source ↗
Figure 4
Figure 4. Figure 4: summarizes the four steps of the empirical strategy. First, we convert worker-profile text into embeddings and combine these with structured profile variables (Section 3.1). Second, we estimate quarter-specific demand models that map the four worker-profile blocks to demand (i.e., log quarterly contracts) (Section 3.2). Third, we compute each block’s importance for pre￾dicted demand using Shapley values (S… view at source ↗
Figure 5
Figure 5. Figure 5: 2D Projection of Self-Presentation Embeddings [PITH_FULL_IMAGE:figures/full_fig_p015_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Observed Demand Event Study 2021Q1 2021Q3 2022Q1 2022Q3 2023Q1 2023Q3 2024Q1 2024Q3 2025Q1 2025Q3 2026Q1 0.20 0.15 0.10 0.05 0.00 0.05 Change in log demand Notes: The dashed vertical line marks the first post-AI quarter. Intervals are percentile bootstrap intervals from the worker-level bootstrap. 4.2 Block Importance: Human Capital Signals and Price We now address our main research question of whether gen… view at source ↗
Figure 7
Figure 7. Figure 7: Block Importance by AI Exposure 2021Q1 2021Q3 2022Q1 2022Q3 2023Q1 2023Q3 2024Q1 2024Q3 2025Q1 2025Q3 2026Q1 0.06 0.04 0.02 0.00 0.02 0.04 Self-Presentation 2021Q1 2021Q3 2022Q1 2022Q3 2023Q1 2023Q3 2024Q1 2024Q3 2025Q1 2025Q3 2026Q1 0.06 0.04 0.02 0.00 0.02 0.04 Credentials 2021Q1 2021Q3 2022Q1 2022Q3 2023Q1 2023Q3 2024Q1 2024Q3 2025Q1 2025Q3 2026Q1 0.06 0.04 0.02 0.00 0.02 0.04 Reputation 2021Q1 2021Q3 2… view at source ↗
Figure 8
Figure 8. Figure 8: Raw Demand Levels by Human Capital and Price Groups [PITH_FULL_IMAGE:figures/full_fig_p026_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Demand Gap Compression by AI Exposure 2021Q1 2021Q3 2022Q1 2022Q3 2023Q1 2023Q3 2024Q1 2024Q3 2025Q1 2025Q3 2026Q1 0.20 0.15 0.10 0.05 0.00 0.05 Change in demand gap 2021Q1 2021Q3 2022Q1 2022Q3 2023Q1 2023Q3 2024Q1 2024Q3 2025Q1 2025Q3 2026Q1 0.20 0.15 0.10 0.05 0.00 0.05 Notes: The left panel plots the event-study coefficient corresponding to TopHC i × Ei, where TopHC i = 1 indicates the worker is in the … view at source ↗
Figure 10
Figure 10. Figure 10: Predictive Performance Across Models and Embeddings [PITH_FULL_IMAGE:figures/full_fig_p035_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Cross-Sectional and Temporal Variability in Prices [PITH_FULL_IMAGE:figures/full_fig_p039_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Permutation Test for Block Importance 0 100 200 300 400 500 600 Count p = 0:020 Self-presentation p = 0:011 Credentials p = 0:056 Reputation p = 0:061 Price 0.05 0.00 0.05 Coefficient 0 100 200 300 400 500 600 Count p = 0:025 0.05 0.00 0.05 Coefficient p = 0:021 0.05 0.00 0.05 Coefficient p = 0:111 0.05 0.00 0.05 Coefficient p = 0:059 Notes: Histograms plot coefficients from the permutation tests. The top… view at source ↗
Figure 13
Figure 13. Figure 13: Permutation Test for Demand Gaps 0 100 200 300 400 Post Permutation count p = 0:348 Human Capital p = 0:406 Price 0.2 0.0 0.2 Coefficient 0 100 200 300 400 LatePost Permutation count p = 0:167 0.2 0.0 0.2 Coefficient p = 0:124 Notes: Histograms plot pooled triple-difference coefficients from the exposure-permutation exercise. The top row reports the baseline post-period specification; the bottom row repor… view at source ↗
Figure 14
Figure 14. Figure 14: Distribution of AI Exposure Scores 0.0 0.2 0.4 0.6 0.8 1.0 AI Exposure Score 0 2000 4000 6000 8000 10000 Count Workers 0.0 0.2 0.4 0.6 0.8 1.0 AI Exposure Score 0 2 4 6 8 10 12 14 16 Count Subcategories Notes: The worker distribution is based on each worker’s primary selected Upwork subcategory; its mean is 0.252 and median is 0.219. The subcategory distribution weights each of the 102 represented Upwork … view at source ↗
read the original abstract

Has generative AI changed how labor markets value human capital? We study this question using data from Upwork, a large online labor market. Representing worker profiles with high-dimensional text embeddings, we compute the importance of human capital information and price in predicting labor demand, and incorporate these measures into a difference-in-differences design around the release of ChatGPT. We find that in more AI-exposed job categories, the importance of human capital declines and the importance of price rises, suggesting a commoditization effect of AI on labor. Two additional findings support commoditization as a mechanism: The demand premium enjoyed by workers with strong human capital declines in more AI-exposed categories, and demand reallocates toward lower-priced workers. Our results have implications for the design of online labor markets, workers' incentives to invest in human capital, and labor welfare.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper claims that generative AI commoditizes labor. Using high-dimensional text embeddings of Upwork worker profiles to measure the predictive importance of human capital attributes versus price, the authors implement a difference-in-differences design around the November 2022 release of ChatGPT. They report that, in more AI-exposed job categories, the importance of human capital information declines while the importance of price rises. Two supporting patterns are presented: the demand premium for workers with strong human capital falls in exposed categories, and demand shifts toward lower-priced workers.

Significance. If the identification and measurement strategy are valid, the results would provide direct evidence that generative AI reduces the market return to human capital signals and increases price sensitivity in online labor markets. This would carry implications for workers' incentives to invest in skills, the design of gig platforms, and aggregate labor welfare. The embedding-based approach to quantifying attribute importance in demand prediction is a methodological contribution that could be applied more broadly.

major comments (3)
  1. [Empirical Strategy] The central DiD claim requires that, absent ChatGPT, trends in the embedding-derived importance of human capital and price would have been parallel across AI-exposed and non-exposed job categories. The abstract and methods description provide no pre-trend evidence, event-study plots, or placebo tests on pre-2022 data; without these, the parallel-trends assumption remains unverified and the causal interpretation of the post-ChatGPT divergence is not supported.
  2. [Data and Measurement] The exposure classification of job categories must be exogenous to other post-2022 demand shifts (e.g., remote-work changes or skill-biased technical change). The manuscript does not report robustness checks using alternative exposure measures, falsification tests on non-AI technologies, or balance checks on observables that might correlate with both exposure and the outcome trends.
  3. [Measurement of Human Capital and Price Importance] The construction of the importance measures from text embeddings is load-bearing for the commoditization interpretation. Details are needed on how the embeddings are trained or fine-tuned, how importance is extracted (e.g., via feature ablation, SHAP values, or coefficient magnitudes in the demand prediction model), and whether these measures are validated against human-coded profiles or out-of-sample predictive accuracy.
minor comments (2)
  1. [Abstract] The abstract refers to 'two additional findings' supporting the mechanism; these should be presented with the same level of detail as the main DiD results, including coefficient magnitudes and standard errors.
  2. [Empirical Strategy] Clarify the exact timing of the ChatGPT release used for the post-period indicator and whether any anticipation or staggered rollout is accounted for.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the detailed and constructive comments. We address each major point below and indicate the revisions we will make to strengthen the identification, measurement, and robustness of the results.

read point-by-point responses

  1. Referee: [Empirical Strategy] The central DiD claim requires that, absent ChatGPT, trends in the embedding-derived importance of human capital and price would have been parallel across AI-exposed and non-exposed job categories. The abstract and methods description provide no pre-trend evidence, event-study plots, or placebo tests on pre-2022 data; without these, the parallel-trends assumption remains unverified and the causal interpretation of the post-ChatGPT divergence is not supported.

    Authors: We agree that explicit verification of parallel trends is essential for causal claims. The current draft emphasizes the post-ChatGPT divergence but does not include pre-trend diagnostics in the main text. In the revision we will add event-study plots with leads and lags around November 2022, placebo tests on pre-2022 data using the same embedding-based importance measures, and a formal test for differential pre-trends. These additions will be placed in a new subsection of the empirical strategy. revision: yes

  2. Referee: [Data and Measurement] The exposure classification of job categories must be exogenous to other post-2022 demand shifts (e.g., remote-work changes or skill-biased technical change). The manuscript does not report robustness checks using alternative exposure measures, falsification tests on non-AI technologies, or balance checks on observables that might correlate with both exposure and the outcome trends.

    Authors: We recognize that the exposure measure, while based on pre-ChatGPT task descriptions, could be correlated with other contemporaneous shocks. In the revised manuscript we will (i) report results using two alternative exposure classifications (one based on an independent LLM coding of task substitutability and one based on occupational exposure scores from prior literature), (ii) conduct falsification tests replacing ChatGPT with earlier non-generative AI milestones, and (iii) present balance tables and covariate-adjusted specifications to address potential confounding. These checks will be added to the robustness section. revision: yes

  3. Referee: [Measurement of Human Capital and Price Importance] The construction of the importance measures from text embeddings is load-bearing for the commoditization interpretation. Details are needed on how the embeddings are trained or fine-tuned, how importance is extracted (e.g., via feature ablation, SHAP values, or coefficient magnitudes in the demand prediction model), and whether these measures are validated against human-coded profiles or out-of-sample predictive accuracy.

    Authors: We will expand the measurement appendix to document: the exact embedding model and any fine-tuning procedure; the precise method used to extract importance (feature ablation on the demand-prediction model, with results cross-checked via permutation importance); and validation exercises comparing the embedding-derived importance rankings to human coders on a held-out sample of profiles as well as out-of-sample predictive performance metrics. These details were condensed in the original submission for space but will be fully reported to allow replication and assessment of the commoditization interpretation. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical DiD estimates from observational data

full rationale

The paper is a standard empirical study applying difference-in-differences to Upwork labor demand data, using text embeddings to construct importance measures for human capital and price. No equations, fitted parameters, or self-citations are presented as deriving the central commoditization result by construction. The DiD design and embedding-based predictors are data-driven estimates subject to identification assumptions, but the derivation chain does not reduce to self-definition, renaming, or load-bearing self-citation. This matches the default case of a self-contained empirical paper with no circular steps.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The analysis rests on standard econometric assumptions for causal identification and the validity of text embeddings as proxies for human capital; no free parameters or invented entities are described in the abstract.

axioms (2)
  • domain assumption Parallel trends assumption holds between AI-exposed and non-exposed job categories absent the ChatGPT shock
    Required for the DiD design to attribute post-ChatGPT changes to AI exposure rather than other trends.
  • domain assumption Text embeddings accurately capture relevant dimensions of human capital information in worker profiles
    Central to computing the importance of human capital in demand prediction.

pith-pipeline@v0.9.1-grok · 5669 in / 1239 out tokens · 38991 ms · 2026-06-26T11:13:41.512337+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

63 extracted references · 41 canonical work pages

  1. [1]

    , address =

    Becker, Gary S. , address =. Human capital : a theoretical and empirical analysis, with special reference to education / by Gary S. Becker. , year =

  2. [2]

    Schooling, experience, and earnings / Jacob Mincer

    Mincer, Jacob , address =. Schooling, experience, and earnings / Jacob Mincer. , year =

  3. [3]

    The Quarterly Journal of Economics , volume =

    Spence, Michael , title =. The Quarterly Journal of Economics , volume =. 1973 , month =. doi:10.2307/1882010 , url =

    work page doi:10.2307/1882010 1973

  4. [4]

    Stiglitz , journal =

    Joseph E. Stiglitz , journal =. The Theory of ``Screening," Education, and the Distribution of Income , urldate =

  5. [5]

    and Pierret, Charles R

    Altonji, Joseph G. and Pierret, Charles R. , title =. The Quarterly Journal of Economics , volume =. 2001 , month =. doi:10.1162/003355301556329 , url =

    work page doi:10.1162/003355301556329 2001

  6. [6]

    and Levy, Frank and Murnane, Richard J

    Autor, David H. and Levy, Frank and Murnane, Richard J. , title =. The Quarterly Journal of Economics , volume =. 2003 , month =. doi:10.1162/003355303322552801 , url =

    work page doi:10.1162/003355303322552801 2003

  7. [7]

    Skills, Tasks and Technologies: Implications for Employment and Earnings , editor =

    Daron Acemoglu and David Autor , keywords =. Skills, Tasks and Technologies: Implications for Employment and Earnings , editor =. 2011 , issn =. doi:https://doi.org/10.1016/S0169-7218(11)02410-5 , url =

    work page doi:10.1016/s0169-7218(11)02410-5 2011

  8. [8]

    Journal of Economic Perspectives , Volume =

    Acemoglu, Daron and Restrepo, Pascual , Title =. Journal of Economic Perspectives , Volume =. 2019 , Month =. doi:10.1257/jep.33.2.3 , URL =

    work page doi:10.1257/jep.33.2.3 2019

  9. [9]

    Journal of Labor Economics , volume =

    Derek Neal , title =. Journal of Labor Economics , volume =. 1995 , doi =. https://www.journals.uchicago.edu/doi/pdf/10.1086/298388 , abstract =

    work page doi:10.1086/298388 1995

  10. [10]

    Lazear , title =

    Edward P. Lazear , title =. Journal of Political Economy , volume =. 2009 , doi =. https://www.journals.uchicago.edu/doi/pdf/10.1086/648671 , abstract =

    work page doi:10.1086/648671 2009

  11. [11]

    American Economic Review , Volume =

    Pallais, Amanda , Title =. American Economic Review , Volume =. 2014 , Month =. doi:10.1257/aer.104.11.3565 , URL =

    work page doi:10.1257/aer.104.11.3565 2014

  12. [12]

    and Thomas, Catherine , title =

    Stanton, Christopher T. and Thomas, Catherine , title =. The Review of Economic Studies , volume =. 2016 , month =. doi:10.1093/restud/rdv042 , url =

    work page doi:10.1093/restud/rdv042 2016

  13. [13]

    Management Science , volume =

    Lin, Mingfeng and Liu, Yong and Viswanathan, Siva , title =. Management Science , volume =. 2018 , doi =. https://doi.org/10.1287/mnsc.2016.2594 , abstract =

    work page doi:10.1287/mnsc.2016.2594 2018

  14. [14]

    , title =

    Kokkodis, Marios and Ipeirotis, Panagiotis G. , title =. Management Science , volume =. 2016 , doi =. https://doi.org/10.1287/mnsc.2015.2217 , abstract =

    work page doi:10.1287/mnsc.2015.2217 2016

  15. [15]

    Information Systems Research , volume =

    Kokkodis, Marios , title =. Information Systems Research , volume =. 2023 , doi =. https://doi.org/10.1287/isre.2022.1177 , abstract =

    work page doi:10.1287/isre.2022.1177 2023

  16. [16]

    Management Science , volume =

    Kokkodis, Marios and Ransbotham, Sam , title =. Management Science , volume =. 2023 , doi =. https://doi.org/10.1287/mnsc.2022.4426 , abstract =

    work page doi:10.1287/mnsc.2022.4426 2023

  17. [17]

    Journal of Labor Economics , volume =

    Ioana Marinescu and Ronald Wolthoff , title =. Journal of Labor Economics , volume =. 2020 , doi =. https://www.journals.uchicago.edu/doi/pdf/10.1086/705903 , abstract =

    work page doi:10.1086/705903 2020

  18. [18]

    Skill Spanning in the Online Labor Market: A Double-Edged Sword? , volume =

    Fu, Yan and Feng, Juan and Ye, Qiang , address =. Skill Spanning in the Online Labor Market: A Double-Edged Sword? , volume =. Journal of the Association for Information Systems , keywords =. 2022 , abstract =

    2022

  19. [19]

    Online Labor Markets

    Horton, John J. Online Labor Markets. Internet and Network Economics. 2010

    2010

  20. [20]

    Horton , title =

    John J. Horton , title =. Journal of Labor Economics , volume =. 2017 , doi =. https://www.journals.uchicago.edu/doi/pdf/10.1086/689213 , abstract =

    work page doi:10.1086/689213 2017

  21. [21]

    Management Science , volume =

    Wiles, Emma and Munyikwa, Zanele and Horton, John , title =. Management Science , volume =. 2025 , doi =. https://doi.org/10.1287/mnsc.2024.04528 , abstract =

    work page doi:10.1287/mnsc.2024.04528 2025

  22. [22]

    2022 , month = nov, url =

    Introducing. 2022 , month = nov, url =

    2022

  23. [23]

    Cowgill, Bo and Hern\'. Does. Management Science , year =. doi:10.1287/mnsc.2024.07027 , URL =. https://doi.org/10.1287/mnsc.2024.07027 , abstract =

    work page doi:10.1287/mnsc.2024.07027 2024

  24. [24]

    Signaling in the Age of

    Jingyi Cui and Gabriel Dias and Justin Ye , year=. Signaling in the Age of. 2509.25054 , archivePrefix=

    arXiv

  25. [25]

    Making Talk Cheap: Generative

    Anais Galdin and Jesse Silbert , year=. Making Talk Cheap: Generative. 2511.08785 , archivePrefix=

    arXiv

  26. [26]

    , title =

    Wiles, Emma and Horton, John J. , title =. 2026 , note =

    2026

  27. [27]

    Science , volume =

    Tyna Eloundou and Sam Manning and Pamela Mishkin and Daniel Rock , title =. Science , volume =. 2024 , doi =. https://www.science.org/doi/pdf/10.1126/science.adj0998 , abstract =

    work page doi:10.1126/science.adj0998 2024

  28. [28]

    2026 , url =

    Massenkoff, Maxim and McCrory, Peter , title =. 2026 , url =

    2026

  29. [29]

    Strategic Management Journal , volume =

    Felten, Edward and Raj, Manav and Seamans, Robert , title =. Strategic Management Journal , volume =. doi:https://doi.org/10.1002/smj.3286 , url =. https://sms.onlinelibrary.wiley.com/doi/pdf/10.1002/smj.3286 , abstract =

    work page doi:10.1002/smj.3286

  30. [30]

    Journal of Labor Economics , volume =

    Daron Acemoglu and David Autor and Jonathon Hazell and Pascual Restrepo , title =. Journal of Labor Economics , volume =. 2022 , doi =. https://www.journals.uchicago.edu/doi/pdf/10.1086/718327 , abstract =

    work page doi:10.1086/718327 2022

  31. [31]

    Economic Policy , volume =

    Acemoglu, Daron , title =. Economic Policy , volume =. 2025 , month =. doi:10.1093/epolic/eiae042 , url =

    work page doi:10.1093/epolic/eiae042 2025

  32. [32]

    The Quarterly Journal of Economics , volume =

    Brynjolfsson, Erik and Li, Danielle and Raymond, Lindsey , title =. The Quarterly Journal of Economics , volume =. 2025 , month =. doi:10.1093/qje/qjae044 , url =

    work page doi:10.1093/qje/qjae044 2025

  33. [33]

    Organization Science , volume =

    Hui, Xiang and Reshef, Oren and Zhou, Luofeng , title =. Organization Science , volume =. 2024 , doi =. https://doi.org/10.1287/orsc.2023.18441 , abstract =

    work page doi:10.1287/orsc.2023.18441 2024

  34. [34]

    Management Science , volume =

    Demirci, Ozge and Hannane, Jonas and Zhu, Xinrong , title =. Management Science , volume =. 2025 , doi =. https://doi.org/10.1287/mnsc.2024.05420 , abstract =

    work page doi:10.1287/mnsc.2024.05420 2025

  35. [35]

    Science , volume =

    Shakked Noy and Whitney Zhang , title =. Science , volume =. 2023 , doi =. https://www.science.org/doi/pdf/10.1126/science.adh2586 , abstract =

    work page doi:10.1126/science.adh2586 2023

  36. [36]

    The Impact of

    Sida Peng and Eirini Kalliamvakou and Peter Cihon and Mert Demirer , year=. The Impact of. 2302.06590 , archivePrefix=

    Pith/arXiv arXiv

  37. [37]

    and Rajendran, Saran and Krayer, Lisa and Candelon, Fran

    Dell’Acqua, Fabrizio and McFowland, Edward and Mollick, Ethan and Lifshitz, Hila and Kellogg, Katherine C. and Rajendran, Saran and Krayer, Lisa and Candelon, Fran. Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of Artificial Intelligence on Knowledge Worker Productivity and Quality , journal =. 2026 , doi =. http...

    work page doi:10.1287/orsc.2025.21838 2026

  38. [38]

    and Prieur, Cl\'

    Owen, Art B. and Prieur, Cl\'. On Shapley Value for Measuring Importance of Dependent Inputs , journal =. 2017 , doi =. https://doi.org/10.1137/16M1097717 , abstract =

    work page doi:10.1137/16m1097717 2017

  39. [39]

    and Lundberg, Scott and Lee, Su-In , title =

    Covert, Ian C. and Lundberg, Scott and Lee, Su-In , title =. Proceedings of the 34th International Conference on Neural Information Processing Systems , articleno =. 2020 , isbn =

    2020

  40. [40]

    Applied Stochastic Models in Business and Industry , volume =

    Lipovetsky, Stan and Conklin, Michael , title =. Applied Stochastic Models in Business and Industry , volume =. doi:https://doi.org/10.1002/asmb.446 , url =. https://onlinelibrary.wiley.com/doi/pdf/10.1002/asmb.446 , abstract =

    work page doi:10.1002/asmb.446

  41. [41]

    Electronic Journal of Statistics , publisher =

    Frank Huettner and Marco Sunder , title =. Electronic Journal of Statistics , publisher =. 2012 , doi =

    2012

  42. [42]

    and Lee, Su-In , title =

    Lundberg, Scott M. and Lee, Su-In , title =. Proceedings of the 31st International Conference on Neural Information Processing Systems , pages =. 2017 , isbn =

    2017

  43. [43]

    http://www.nber.org/papers/w32117

    Callaway, Brantly and Goodman-Bacon, Andrew and Sant'Anna, Pedro H. C. Difference-in-differences with a Continuous Treatment. 2024. doi:10.3386/w32117 , URL = "http://www.nber.org/papers/w32117", abstract =

    work page doi:10.3386/w32117 2024

  44. [44]

    Sant’Anna , keywords =

    Brantly Callaway and Pedro H.C. Sant’Anna , keywords =. Difference-in-Differences with multiple time periods , journal =. 2021 , note =. doi:https://doi.org/10.1016/j.jeconom.2020.12.001 , url =

    work page doi:10.1016/j.jeconom.2020.12.001 2021

  45. [45]

    and Lange, Fabian , title =

    Kahn, Lisa B. and Lange, Fabian , title =. The Review of Economic Studies , volume =. 2014 , month =. doi:10.1093/restud/rdu021 , url =

    work page doi:10.1093/restud/rdu021 2014

  46. [46]

    Crawford , title =

    Joel Sobel AND Vincent P. Crawford , title =. Econometrica , volume =

  47. [47]

    Journal of Political Economy , volume =

    Sherwin Rosen , title =. Journal of Political Economy , volume =. 1974 , doi =

    1974

  48. [48]

    , Title =

    Petrongolo, Barbara and Pissarides, Christopher A. , Title =. Journal of Economic Literature , Volume =. 2001 , Month =. doi:10.1257/jel.39.2.390 , URL =

    work page doi:10.1257/jel.39.2.390 2001

  49. [49]

    The Quarterly Journal of Economics , volume =

    Kleinberg, Jon and Lakkaraju, Himabindu and Leskovec, Jure and Ludwig, Jens and Mullainathan, Sendhil , title =. The Quarterly Journal of Economics , volume =. 2018 , month =. doi:10.1093/qje/qjx032 , url =

    work page doi:10.1093/qje/qjx032 2018

  50. [50]

    The Quarterly Journal of Economics , volume =

    Abadie, Alberto and Athey, Susan and Imbens, Guido W and Wooldridge, Jeffrey M , title =. The Quarterly Journal of Economics , volume =. 2023 , month =. doi:10.1093/qje/qjac038 , url =

    work page doi:10.1093/qje/qjac038 2023

  51. [51]

    Proceedings of the 27th International Conference on Neural Information Processing Systems - Volume 2 , pages =

    Mikolov, Tomas and Sutskever, Ilya and Chen, Kai and Corrado, Greg and Dean, Jeffrey , title =. Proceedings of the 27th International Conference on Neural Information Processing Systems - Volume 2 , pages =. 2013 , publisher =

    2013

  52. [52]

    Proceedings of the 34th International Conference on Neural Information Processing Systems , articleno =

    Wang, Wenhui and Wei, Furu and Dong, Li and Bao, Hangbo and Yang, Nan and Zhou, Ming , title =. Proceedings of the 34th International Conference on Neural Information Processing Systems , articleno =. 2020 , isbn =

    2020

  53. [53]

    2024 , eprint=

    Arctic-Embed 2.0: Multilingual Retrieval Without Compromise , author=. 2024 , eprint=

    2024

  54. [54]

    Journal of Machine Learning Research , year =

    Laurens van der Maaten and Geoffrey Hinton , title =. Journal of Machine Learning Research , year =

  55. [55]

    Hoerl and Robert W

    Arthur E. Hoerl and Robert W. Kennard , title =. Technometrics , volume =. 1970 , publisher =. doi:10.1080/00401706.1970.10488634 , URL =

    work page doi:10.1080/00401706.1970.10488634 1970

  56. [56]

    Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , pages =

    Chen, Tianqi and Guestrin, Carlos , title =. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , pages =. 2016 , isbn =. doi:10.1145/2939672.2939785 , abstract =

    work page doi:10.1145/2939672.2939785 2016

  57. [57]

    Journal of Political Economy , volume =

    Paul Milgrom and John Roberts , title =. Journal of Political Economy , volume =. 1986 , doi =. https://www.journals.uchicago.edu/doi/pdf/10.1086/261408 , abstract =

    work page doi:10.1086/261408 1986

  58. [58]

    The Commoditization of Labor , note =

    Fukui, Masao and Nakamura, Emi and Steinsson, J. The Commoditization of Labor , note =. 2026 , url =

    2026

  59. [59]

    American Economic Review: Insights , Volume =

    Roth, Jonathan , Title =. American Economic Review: Insights , Volume =. 2022 , Month =. doi:10.1257/aeri.20210236 , URL =

    work page doi:10.1257/aeri.20210236 2022

  60. [60]

    P. J. Bickel and F. Götze and W. R. van Zwet , journal =. RESAMPLING FEWER THAN n OBSERVATIONS: GAINS, LOSSES, AND REMEDIES FOR LOSSES , urldate =

  61. [61]

    Blei , title =

    Keyon Vafa and Susan Athey and David M. Blei , title =. Proceedings of the National Academy of Sciences , volume =. 2025 , doi =. https://www.pnas.org/doi/pdf/10.1073/pnas.2427298122 , abstract =

    work page doi:10.1073/pnas.2427298122 2025

  62. [62]

    2025 , date =

    Xu, Jiannan and Li, Gujie and Jiang, Jane Yi , title =. 2025 , date =. doi:10.2139/ssrn.5417394 , url =

    work page doi:10.2139/ssrn.5417394 2025

  63. [63]

    Tibshirani and Larry Wasserman , title =

    Jing Lei and Max G’Sell and Alessandro Rinaldo and Ryan J. Tibshirani and Larry Wasserman , title =. Journal of the American Statistical Association , volume =. 2018 , publisher =. doi:10.1080/01621459.2017.1307116 , URL =. https://doi.org/10.1080/01621459.2017.1307116 , abstract =

    work page doi:10.1080/01621459.2017.1307116 2018