pith. sign in

arxiv: 2607.01101 · v1 · pith:7YRXJSO2new · submitted 2026-07-01 · 💰 econ.GN · q-fin.EC

The Economic Benefits and Costs of AI and Policies to Mitigate AI's Impact on Inequality

Matthew O. Jackson , Zafer Kanik This is my paper

Pith reviewed 2026-07-02 02:46 UTC · model grok-4.3

classification 💰 econ.GN q-fin.EC
keywords AI productivitylabor reallocationwage inequalityAI monopolyeconomic policyPareto improvementslabor categoriessubstitution elasticity
0
0 comments X

The pith

AI productivity gains raise wages for workers essential to building AI faster than GDP while lowering wages for workers it replaces.

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

The paper models how rising AI productivity drives labor reallocation across three distinct categories and alters absolute and relative wages. Labor essential for AI production sees wages grow faster than overall GDP. Labor substituted by AI experiences both absolute and relative wage declines. Labor used only in final goods production tracks GDP growth. The effects and the optimal policies to spread benefits differ sharply depending on whether AI production occurs under competition or monopoly.

Core claim

Improvements in AI productivity trigger labor reallocation and changes in absolute and relative wages for different types of labor. Wages of labor that is essential for building AI increase faster than overall GDP. Wages of labor that is substituted for by AI decrease in both absolute and relative terms. Wages of labor that is used only in final goods production and is not displaced by AI increase in line with overall GDP. Monopoly production of AI restricts its deployment, slowing the transition and impact of AI. Optimal tax and regulatory policies that achieve Pareto-improvements differ depending on whether there is competition in AI production.

What carries the argument

A partition of labor into three fixed categories with distinct elasticities of substitution to AI, applied under both competitive and monopolistic AI production.

If this is right

  • Wages of AI-essential labor grow faster than GDP.
  • Wages of AI-substituted labor fall in both absolute and relative terms.
  • Wages of non-displaced final-goods labor rise in line with GDP.
  • Monopoly AI production slows deployment and limits overall economic effects relative to competition.
  • Pareto-improving tax and regulatory policies take different forms depending on whether AI production is competitive or monopolistic.

Where Pith is reading between the lines

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

  • If workers can readily switch between labor categories, the predicted wage divergence would shrink and policy needs would change.
  • The framework implies that inequality from AI may be more persistent under monopoly production, suggesting competition policy as a complement to redistribution.
  • Extending the model to allow gradual labor mobility or continuous substitution elasticities could test how robust the policy differences remain.
  • The results point to monitoring sector-specific wage trends as AI advances to detect early signs of the reallocation effects.

Load-bearing premise

Labor remains divided into three fixed, non-overlapping categories that cannot switch or exhibit continuous substitution patterns with AI.

What would settle it

Wage data showing no faster growth for AI-building labor than GDP and no absolute decline for substituted labor as AI productivity rises would falsify the predicted divergence.

Figures

Figures reproduced from arXiv: 2607.01101 by Matthew O. Jackson, Zafer Kanik.

Figure 1
Figure 1. Figure 1: AI Adoption and Equilibrium Wages Notes: This figure displays wages as AI productivity (Aa) increases from Aa = 0 to Aa = 6. The three panels show wages for substitutable (wS), complementary (wC ), and final goods specific (wF ) labor. Two AI technologies are compared: the solid blue line depicts the case where AI is produced exclusively by complementary labor (α S a = 0) resulting in an endless transition… view at source ↗
Figure 2
Figure 2. Figure 2: AI Adoption and Wage Inequality the changes in wF /wS and wC/wF depends on whether Λ exceeds unity: when Λ > 1, final goods specific labor gains faster relative to substitutable labor than complementary labor gains relative to final goods specific labor. Lastly, when α S a = 0 (solid lines), wC/wS and wF /wS grow without bound as the economy remains in endless transition, while wC/wF rises toward the finit… view at source ↗
Figure 3
Figure 3. Figure 3: The AI Expenditure Share in Final Goods Sector and the Aggregate Output [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Pareto Improving Tax Policies: An economy starts at some productivity [PITH_FULL_IMAGE:figures/full_fig_p020_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: monopoly Profit vs Substitutable-Worker Deficit [PITH_FULL_IMAGE:figures/full_fig_p027_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Output Levels and Workers’ Gains and Losses under Competition vs. Monopoly [PITH_FULL_IMAGE:figures/full_fig_p028_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: After-Tax Complementary Wage under Minimal S-Worker Protection [PITH_FULL_IMAGE:figures/full_fig_p029_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: AI-in-AI: Adoption, Growth, and Inequality [PITH_FULL_IMAGE:figures/full_fig_p032_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: AI Productivity and General Equilibrium Wage Dynamics [PITH_FULL_IMAGE:figures/full_fig_p082_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Elasticities of Wage Inequality with Respect to AI Productivity [PITH_FULL_IMAGE:figures/full_fig_p084_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Technological Determinants of Transition Duration ( [PITH_FULL_IMAGE:figures/full_fig_p088_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: GE-Monopoly, Cost-Minimizing Monopoly, and Competition [PITH_FULL_IMAGE:figures/full_fig_p103_12.png] view at source ↗
read the original abstract

We examine the economic impact of increasingly productive AI and policies that spread its benefits across the economy. Improvements in AI productivity trigger labor reallocation and changes in absolute and relative wages for different types of labor. Wages of labor that is essential for building AI increase faster than overall GDP. Wages of labor that is substituted for by AI decrease in both absolute and relative terms. Wages of labor that is used only in final goods production and is not displaced by AI increase in line with overall GDP. We contrast the impact of productivity gains depending on whether AI production is competitive or monopolistic. Monopoly production of AI restricts its deployment, slowing the transition and impact of AI. Optimal tax and regulatory policies that achieve Pareto-improvements differ depending on whether there is competition in AI production.

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

2 major / 1 minor

Summary. The manuscript develops a theoretical general equilibrium model partitioning labor into three fixed categories (AI-essential, AI-substitutable, and final-goods only) to examine how AI productivity improvements affect labor reallocation, absolute and relative wages, and inequality. It claims that AI-essential labor wages grow faster than GDP, substitutable labor wages decline in both absolute and relative terms, and final-goods labor wages track GDP. The model contrasts outcomes under competitive versus monopolistic AI production and derives optimal tax and regulatory policies that differ by market structure.

Significance. If the central results hold under the stated assumptions, the paper provides a clear framework for analyzing AI-driven distributional effects and the role of market structure in AI deployment. The explicit comparison of competitive and monopolistic AI production is a useful distinction for policy analysis.

major comments (2)
  1. [Section 2] Section 2 and the equilibrium conditions: the model treats the three labor categories as fixed and non-overlapping with category-specific substitution elasticities and no cross-category mobility. This assumption is load-bearing for the wage divergence predictions; the absolute and relative wage declines for the substitutable group, as well as the policy distinctions, follow directly from the immutability of categories. If even modest worker reallocation across categories were permitted, the predicted effects of productivity shocks would not follow in the same form.
  2. [Equilibrium conditions] Equilibrium conditions: the claims that wages of AI-essential labor increase faster than GDP and substitutable labor wages decrease rely on the fixed partitioning and distinct elasticities. The manuscript does not examine robustness to a continuum of tasks or endogenous category choice, which would alter the reallocation and wage results.
minor comments (1)
  1. The abstract is concise, but the introduction should more explicitly flag the fixed labor categories as a modeling choice rather than an empirical claim.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments highlighting the role of our labor category assumptions. We respond to each major comment below.

read point-by-point responses

  1. Referee: [Section 2] Section 2 and the equilibrium conditions: the model treats the three labor categories as fixed and non-overlapping with category-specific substitution elasticities and no cross-category mobility. This assumption is load-bearing for the wage divergence predictions; the absolute and relative wage declines for the substitutable group, as well as the policy distinctions, follow directly from the immutability of categories. If even modest worker reallocation across categories were permitted, the predicted effects of productivity shocks would not follow in the same form.

    Authors: The fixed partitioning into three non-overlapping categories is an intentional modeling choice that isolates the distinct effects of AI productivity on essential, substitutable, and final-goods labor, consistent with observed differences in skill requirements and limited short-run mobility. This structure generates the wage divergence results under the stated substitution elasticities. We agree that permitting cross-category reallocation would dampen or alter these effects. We will revise Section 2 to more explicitly state this assumption, its motivation, and the conditions under which the predictions hold. revision: partial

  2. Referee: [Equilibrium conditions] Equilibrium conditions: the claims that wages of AI-essential labor increase faster than GDP and substitutable labor wages decrease rely on the fixed partitioning and distinct elasticities. The manuscript does not examine robustness to a continuum of tasks or endogenous category choice, which would alter the reallocation and wage results.

    Authors: The equilibrium conditions and wage claims are derived directly from the fixed categories and category-specific elasticities. Extending the framework to a continuum of tasks or endogenous category choice would constitute a different model and is outside the paper's scope, which focuses on the implications of the three-type partition. We will add a brief discussion in the conclusion noting this modeling choice and the conditional nature of the results. revision: partial

Circularity Check

0 steps flagged

No significant circularity; standard theoretical model from explicit assumptions

full rationale

The paper is a theoretical economic model deriving labor reallocation and wage effects from market-clearing equilibrium conditions given three fixed labor categories as primitives. No equations reduce by construction to fitted parameters or self-definitions; no self-citation chains justify core premises; no ansatzes or renamings of known results are presented as derivations. The fixed categories are stated assumptions, not outputs, and results follow from solving the model under productivity shocks. This matches the default case of a self-contained theoretical exercise with independent content.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Abstract-only review; the model implicitly rests on standard neoclassical assumptions about labor markets and production that are not enumerated here.

axioms (2)
  • domain assumption Labor can be partitioned into three distinct categories with fixed substitution patterns relative to AI.
    Invoked to generate the three different wage trajectories described in the abstract.
  • standard math Markets clear and wages adjust to equate supply and demand in each labor category.
    Standard general-equilibrium assumption required for the reallocation and wage-change results.

pith-pipeline@v0.9.1-grok · 5659 in / 1304 out tokens · 47902 ms · 2026-07-02T02:46:38.560082+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

72 extracted references

  1. [1]

    Strategic management journal , volume=

    Occupational, industry, and geographic exposure to artificial intelligence: A novel dataset and its potential uses , author=. Strategic management journal , volume=. 2021 , publisher=

    2021

  2. [2]

    The Quarterly Journal of Economics , volume =

    Acemoglu, Daron , title =. The Quarterly Journal of Economics , volume =

  3. [3]

    2025 , note=

    Canaries in the Coal Mine?: Six Facts about the Recent Employment Effects of Artificial Intelligence , author=. 2025 , note=

    2025

  4. [4]

    preprint Bocconi , year =

    Anelli, Massimo and Morelli, Massimo and Pappalettera, Marvin , title =. preprint Bocconi , year =

  5. [5]

    The Wall Street Journal , year =

    Higgins, Tim , title =. The Wall Street Journal , year =

  6. [6]

    The Wall Street Journal , year =

    Hagey, Keach , title =. The Wall Street Journal , year =

  7. [7]

    The Review of Economic Studies , volume=

    Growth accounting with intermediate inputs , author=. The Review of Economic Studies , volume=. 1978 , doi=

    1978

  8. [8]

    Journal of Economic Literature , volume =

    Acemoglu, Daron , title =. Journal of Economic Literature , volume =

  9. [9]

    The Review of Economic Studies , volume=

    Directed technical change , author=. The Review of Economic Studies , volume=. 2002b , publisher=

  10. [10]

    Handbook of Labor Economics , volume =

    Acemoglu, Daron and Autor, David , title =. Handbook of Labor Economics , volume =

  11. [11]

    American Economic Review , volume =

    Acemoglu, Daron and Restrepo, Pascual , title =. American Economic Review , volume =

  12. [12]

    Journal of Political Economy , volume =

    Acemoglu, Daron and Restrepo, Pascual , title =. Journal of Political Economy , volume =

  13. [13]

    The Economics of Artificial Intelligence: An Agenda , pages =

    Acemoglu, Daron and Restrepo, Pascual , title =. The Economics of Artificial Intelligence: An Agenda , pages =

  14. [14]

    Econometrica , volume =

    Acemoglu, Daron and Restrepo, Pascual , title =. Econometrica , volume =

  15. [15]

    and Jones, Charles I

    Aghion, Philippe and Jones, Benjamin F. and Jones, Charles I. , title =

  16. [16]

    2007 , publisher=

    The race between education and technology: The evolution of US educational wage differentials, 1890 to 2005 , author=. 2007 , publisher=

    2005

  17. [17]

    and Dorn, David , title =

    Autor, David H. and Dorn, David , title =. American Economic Review , volume =

  18. [18]

    and Levy, Frank and Murnane, Richard J

    Autor, David H. and Levy, Frank and Murnane, Richard J. , title =. The Quarterly Journal of Economics , volume =

  19. [19]

    , title =

    Autor, David and Goldin, Claudia and Katz, Lawrence F. , title =. AEA Papers and Proceedings , volume =

  20. [20]

    and Katz, Lawrence F

    Autor, David H. and Katz, Lawrence F. and Kearney, Melissa S. , title =. The Review of Economics and Statistics , volume =

  21. [21]

    and Katz, Lawrence F

    Autor, David H. and Katz, Lawrence F. and Krueger, Alan B. , title =. The Quarterly Journal of Economics , volume =

  22. [22]

    Brookings Papers on Economic Activity , pages =

    Autor, David and Salomons, Anna , title =. Brookings Papers on Economic Activity , pages =

  23. [23]

    , title =

    Card, David and DiNardo, John E. , title =. Journal of Labor Economics , volume =

  24. [24]

    and Lewis, Ethan G

    Clemens, Michael A. and Lewis, Ethan G. and Postel, Hannah M. , title =. American Economic Review , volume =

  25. [25]

    The Review of Economics and Statistics , pages =

    Griliches, Zvi , title =. The Review of Economics and Statistics , pages =

  26. [26]

    Journal of Monetary Economics , volume =

    Jaimovich, Nir and Saporta-Eksten, Itay and Siu, Henry and Yedid-Levi, Yaniv , title =. Journal of Monetary Economics , volume =

  27. [27]

    The Quarterly Journal of Economics , volume =

    Lewis, Ethan , title =. The Quarterly Journal of Economics , volume =

  28. [28]

    American Economic Journal: Macroeconomics , volume=

    The rise of the machines: Automation, horizontal innovation, and income inequality , author=. American Economic Journal: Macroeconomics , volume=. 2022 , publisher=

    2022

  29. [29]

    and Kanik, Zafer , title =

    Jackson, Matthew O. and Kanik, Zafer , title =. 2019 , note =

    2019

  30. [30]

    , title =

    Kanik, Zafer and Jackson, Matthew O. , title =. 2026 , note =

    2026

  31. [31]

    and Murphy, Kevin M

    Katz, Lawrence F. and Murphy, Kevin M. , title =. The Quarterly Journal of Economics , volume =

  32. [32]

    Econometrica , volume=

    Uneven growth: automation's impact on income and wealth inequality , author=. Econometrica , volume=. 2022 , publisher=

    2022

  33. [33]

    American Economic Review , volume=

    Wages and employment in the United States and Germany: What explains the differences? , author=. American Economic Review , volume=. 2003 , publisher=

    2003

  34. [34]

    Journal of political Economy , volume=

    Should the personal computer be considered a technological revolution? Evidence from US metropolitan areas , author=. Journal of political Economy , volume=. 2010 , publisher=

    2010

  35. [35]

    , title =

    Goldin, Claudia and Katz, Lawrence F. , title =. The Quarterly Journal of Economics , volume =

  36. [36]

    Katz, L. F. and Autor, D. H. , title =. Handbook of Labor Economics , volume =

  37. [37]

    and Rios-Rull, Jose-Victor and Violante, Giovanni L

    Krusell, Per and Ohanian, Lee E. and Rios-Rull, Jose-Victor and Violante, Giovanni L. , title =. Econometrica , volume =

  38. [38]

    The American Economic Review , volume=

    Long-run implications of investment-specific technological change , author=. The American Economic Review , volume=. 1997 , publisher=

    1997

  39. [39]

    The Quarterly Journal of Economics , volume=

    Productivity and misallocation in general equilibrium , author=. The Quarterly Journal of Economics , volume=. 2020 , publisher=

    2020

  40. [40]

    Econometrica , volume=

    The macroeconomic impact of microeconomic shocks: Beyond Hulten's theorem , author=. Econometrica , volume=. 2019 , publisher=

    2019

  41. [41]

    Carnegie-Rochester conference series on public policy , volume=

    1974 , author=. Carnegie-Rochester conference series on public policy , volume=. 1997 , organization=

    1974

  42. [42]

    Tinbergen, Jan , title =

  43. [43]

    Quarterly Journal of Economics , volume =

    Zeira, Joseph , title =. Quarterly Journal of Economics , volume =

  44. [44]

    Journal of Economic Growth , volume =

    Nakamura, Hideki and Zeira, Joseph , title =. Journal of Economic Growth , volume =. 2024 , doi =

    2024

  45. [45]

    Artificial Intelligence in the Knowledge Economy , journal =

    Ide, Enrique and Talam. Artificial Intelligence in the Knowledge Economy , journal =. 2025 , doi =

    2025

  46. [46]

    The impact of AI on global knowledge work , journal =

    Ide, Enrique and Talam. The impact of AI on global knowledge work , journal =

  47. [47]

    The Turing Valley: How

    Ide, Enrique and Talam. The Turing Valley: How. Working paper , year =

  48. [48]

    2024 , doi =

    Autor, David , title =. 2024 , doi =

    2024

  49. [49]

    The Quarterly Journal of Economics , year =

    Brynjolfsson, Erik and Li, Danielle and Raymond, Lindsey , title =. The Quarterly Journal of Economics , year =

  50. [50]

    Should Robots Be Taxed? , journal =

    Guerreiro, Jo\. Should Robots Be Taxed? , journal =

  51. [51]

    Robots, Trade, and Luddism: A Sufficient Statistic Approach to Optimal Technology Regulation , journal =

    Costinot, Arnaud and Werning, Iv\'. Robots, Trade, and Luddism: A Sufficient Statistic Approach to Optimal Technology Regulation , journal =. 2023 , doi =

    2023

  52. [52]

    CESifo Working Paper No.\ 11084 , year =

    Bastani, Spencer and Waldenstr\". CESifo Working Paper No.\ 11084 , year =

  53. [53]

    , title =

    Bryan, Kevin A. , title =. Journal of Economic Literature , volume =. 2026 , doi =

    2026

  54. [54]

    Working paper , year =

    Althoff, Lukas and Reichardt, Hugo , title =. Working paper , year =

  55. [55]

    2025 , month = may, note =

    Amodei, Dario , title =. 2025 , month = may, note =

    2025

  56. [56]

    Industrial Policy for the Intelligence Age: Ideas to Keep People First , year =

  57. [57]

    , title =

    Korinek, Anton and Stiglitz, Joseph E. , title =. The Economics of Artificial Intelligence: An Agenda , editor =. 2019 , pages =

    2019

  58. [58]

    Journal of the European Economic Association , year =

    Thuemmel, Uwe , title =. Journal of the European Economic Association , year =

  59. [59]

    Economic Policy , year =

    Korinek, Anton and Vipra, Jai , title =. Economic Policy , year =

  60. [60]

    , title =

    Athey, Susan and Scott Morton, Fiona M. , title =. 2025 , note =

    2025

  61. [61]

    , title =

    Korinek, Anton and Lockwood, Lee M. , title =

  62. [62]

    Economic Policy , year =

    Acemoglu, Daron , title =. Economic Policy , year =

  63. [63]

    Journal of Economic Perspectives , year =

    Acemoglu, Daron and Restrepo, Pascual , title =. Journal of Economic Perspectives , year =

  64. [64]

    , title =

    Korinek, Anton and Stiglitz, Joseph E. , title =. 2026 , doi =

    2026

  65. [65]

    and Jones, Charles I

    Aghion, Philippe and Jones, Benjamin F. and Jones, Charles I. , title =. The Economics of Artificial Intelligence: An Agenda , editor =. 2019 , pages =

    2019

  66. [66]

    Oxford Economic Papers , year =

    H\"otte, Kerstin and Theodorakopoulos, Angelos and Koutroumpis, Pantelis , title =. Oxford Economic Papers , year =

  67. [67]

    and Dorn, David , title =

    Autor, David H. and Dorn, David , title =. American Economic Review , year =

  68. [68]

    American Economic Review , year =

    Goos, Maarten and Manning, Alan and Salomons, Anna , title =. American Economic Review , year =

  69. [69]

    Brookings Papers on Economic Activity , year =

    Acemoglu, Daron and Manera, Andrea and Restrepo, Pascual , title =. Brookings Papers on Economic Activity , year =

  70. [70]

    2024 , month =

    Korinek, Anton and Suh, Donghyun , title =. 2024 , month =

    2024

  71. [71]

    The Quarterly Journal of Economics , year =

    Garicano, Luis and Rossi-Hansberg, Esteban , title =. The Quarterly Journal of Economics , year =

  72. [72]

    Annual Review of Economics , year =

    Garicano, Luis and Rossi-Hansberg, Esteban , title =. Annual Review of Economics , year =