Pith. sign in

REVIEW 4 major objections 6 minor 15 references

Artificial General Intelligence and the End of Human Employment: The Need to Renegotiate the Social Contract

T0 review · 4 major / 6 minor · reviewed 2026-08-08 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper argues that AGI will drive human wages to zero and make a new social contract necessary.

desk verdict The paper's central wage-collapse result is mathematically backwards under its own Cobb-Douglas equations, leaving a policy essay that adds little to the existing AGI-and-labor literature. read the letter →

arxiv 2502.07050 v1 pith:K6ORBOHC submitted 2025-02-10 econ.GN q-fin.EC

Pascal Stiefenhofer This is my paper · ORCID
classification econ.GNq-fin.EC
keywords artificialgeneralintelligenceCobb-Douglasproductionfunctionwagecollapseaggregatedemanduniversalbasicincomesocialcontractlabordisplacementwealthconcentration
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to establish that AGI, once it can substitute for human labor, undermines the wage-based economy: human wages fall to zero, income concentrates among AGI and capital owners, and aggregate demand collapses. It uses extended Cobb-Douglas production functions to derive this trajectory and an index of human economic power that decays as AGI labor expands. A sympathetic reader would care because the conclusion is that the social contract must be renegotiated through UBI, cooperative AGI ownership, or progressive AGI capital taxation to avoid economic collapse. The argument is a formalization of a widely discussed risk, not an empirical prediction.

What carries the argument

The machinery is the Cobb-Douglas production function, extended in three versions: AGI as capital ($Y=A(K+K_{AGI})^{\alpha}L^{\beta}$), AGI as labor ($Y=AK^{\alpha}L_1^{\beta_1}L_2^{\beta_2}$), and both AGI labor and AGI capital ($Y=AK^{\alpha}K_{AGI}^{\gamma}L_h^{\beta_1}L_{AGI}^{\beta_2}$). Wages are identified with marginal products, and the limiting argument sends human labor to zero; the normalized power index $P_h=w_hL_h/(w_hL_h+w_{AGI}L_{AGI})$ is then specified to decay exponentially with $L_{AGI}$.

What would settle it

Evaluate the paper's own wage equation (10) at a small positive $L_1$ with $0<\beta_1<1$: the factor $L_1^{\beta_1-1}$ diverges to infinity, so the predicted wage is arbitrarily large, not zero; repeating this calculation for any sequence $L_1\to 0$ would overturn the claimed limit.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is that modeling AGI as either capital or labor inside a Cobb-Douglas production function makes human labor redundant in the long run. Since wages are set by marginal product, the labor term vanishes from the production function once AGI fully replaces humans, and the wage converges to zero; with no wage income, consumption demand collapses even as AGI-driven output expands. The paper also constructs a normalized measure of human economic power, $P_h$, that falls exponentially as the AGI labor share rises, and treats this as a continuous transition from a decentralized human-work economy to a centralized AGI-capital economy.

Load-bearing premise

The load-bearing premise is that the Cobb-Douglas marginal-product formula remains the correct wage rule as human labor shrinks to zero; under that formula the wage blows up instead of vanishing, so the zero-wage conclusion depends on an unstated different production function or on removing human labor from the technology.

Editorial extensions

If this is right

  • If human wages fall to zero, employment-based income cannot support consumption, so the economy faces a Keynesian demand crisis even as production rises.
  • All economic surplus accrues to AGI owners, concentrating wealth and freezing social mobility.
  • Standard income-distribution models, which depend on wage labor, stop describing the economy and must be replaced.
  • Policies such as UBI, public or cooperative AGI ownership, and progressive AGI capital taxation become necessary, not optional, for stability.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same Cobb-Douglas logic, taken literally, points the other way: the stated wage formula diverges as human labor approaches zero, so a consistent model of a post-labor economy needs a different production technology or an explicit assumption that human labor exits the production function entirely.
  • If the zero-wage result is replaced by a high-wage result for the last remaining workers, the political conclusion still holds only if ownership of AGI is concentrated; a testable extension would compare wage shares in AI-intensive sectors against the model's predicted monotone decline.
  • The social-contract framing implies a natural policy comparison: UBI versus capital grants versus public ownership can be evaluated by the same aggregate-demand criterion, which the paper does not perform.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper argues that AGI labor and AGI capital, once integrated into Cobb-Douglas production functions, drive human wages to zero, eliminate human employment, concentrate income among AGI owners, and collapse aggregate demand; it concludes that the social contract must be renegotiated through policies such as UBI, cooperative AGI ownership, and progressive AGI capital taxation. The formal argument proceeds in three models: AGI as capital (Section 2), AGI as labor alongside human labor (Section 3), and AGI as both labor and capital (Section 4). Section 5 introduces a normalized measure of human economic power and postulates an exponential wage decline to describe the transition from decentralized to centralized economic structures. The paper's conclusion is that this transition is mathematically inevitable and that policy intervention is urgent.

Significance. The paper addresses an important and timely question about the economic consequences of AGI and proposes a set of policy responses that are broadly consistent with current debate. If the formal result were valid, the paper would provide a simple demonstration that AGI substitution drives human wages to zero and that aggregate demand collapses. The paper also correctly identifies that the distribution of AGI-owned capital, rather than labor productivity alone, determines the welfare outcome. However, the central theoretical claim is not supported by the model as written: the marginal-product calculations imply the opposite of the wage-collapse conclusion, and the exponential wage decline used later is assumed rather than derived. Because the formal core is mathematically incorrect, the paper cannot substantiate its central claim in its current form.

major comments (4)
  1. [Section 2.2, Eq. (7)] The statement that w → 0 as L → 0 is the reverse of what Eq. (7) implies. For the standard Cobb-Douglas convention β ∈ (0,1), the term L^(β−1) has a negative exponent, so as L → 0+ the marginal product βA(K+K_AGI)^α L^(β−1) diverges to +∞, not to zero. Consequently Eq. (8) does not follow, and the wage-collapse conclusion of Model I is invalid.
  2. [Section 3.1, Eqs. (10) and (13)] The same sign error appears in Model II. In Eq. (10), w1 = β1 A K^α L1^(β1−1) L2^β2; with β1 < 1 the factor L1^(β1−1) diverges as L1 → 0+, so w1 tends to infinity, not zero. Moreover, because L1^β1 multiplies the production function in Eq. (9), taking L1 → 0 drives total output Y to zero rather than to a positive AGI-only outcome. Thus Eq. (13), the paper's central claim that human wages drop to zero, is contradicted by the model's own production function.
  3. [Section 4.1, Eqs. (16) and (20)] The claim that increasing AGI capital K_AGI leads to wh → 0 is not supported by Eq. (16). K_AGI appears only through the positive multiplicative factor K_AGI^γ, so for fixed Lh and LAGI the marginal product of human labor rises with K_AGI rather than falling. To obtain wh → 0 one must take Lh → 0, but then Lh^(β1−1) again diverges for β1 < 1. The limiting argument in Eqs. (18)–(20) therefore does not produce the stated wage collapse.
  4. [Section 5.1, Eqs. (23) and (24)] The exponential wage decline wh = w0 e^(−λ L_AGI) is introduced as an assumption, not derived from Model III. This matters because the earlier marginal-product analysis is supposed to justify the wage collapse; when that analysis fails, Eq. (23) imports the conclusion through an exogenous functional form. In addition, Eq. (24) is internally odd: w∞ is described as 'the asymptotic wage level of AGI labor, typically approaching zero', which conflicts with the idea that AGI wages grow in proportion to AGI capital. Without a derivation of Eq. (23) from the production structure, Section 5's power-decline result is an illustration of an assumed path rather than a model-based prediction.
minor comments (6)
  1. [Title and Abstract] The phrase 'in form of renegotiation the Social Contract' is grammatically incomplete and should be revised to 'in the form of renegotiating the social contract.'
  2. [Introduction, keywords] The keyword 'Resource Missalocation' contains a typo; it should be 'Resource Misallocation.'
  3. [Section 3, Eq. (9)] There is a typo in the text introducing Eq. (9): 'an extende Cobb Douglas pprodiuction function' should read 'an extended Cobb-Douglas production function.'
  4. [Section 5.1, Eq. (24)] The phrase 'typically approaching zero' is unclear; if w∞ is the asymptotic AGI wage level, it should be a positive parameter, not a quantity approaching zero. This ambiguity should be resolved.
  5. [References] References [2] and [10] appear to be the same item and should be merged; several entries are missing complete publication details or arXiv identifiers.
  6. [General] The paper contains numerous formatting and spacing errors, including missing spaces between words in Section 3.2 and elsewhere; a careful copyedit is needed.

Circularity Check

2 steps flagged · score 7.0 of 10

Section 5's exponential decline of human economic power is assumed, not derived: Eq. (23) postulates wh = w0 e^{-lambda L_AGI} and Eq. (27) then algebraically returns that same exponential in Ph, while the normalization Lh + LAGI = 1 forces Ph toward 0 as LAGI goes to 1.

  1. self definitional [Section 5.1, Eqs. (22)-(27)]
    "From our economic model III, human wages decrease as AGI labor increases. We model this decline exponentially by wh = w0e−λLAGI ... Substituting into our power equation yields Ph = [w0e−λLAGI(1−LAGI)] / [w0e−λLAGI(1−LAGI)+w∞(1−e−λLAGI)LAGI]."

    The paper presents Eq. (27) as the derived index of declining human economic power, but the decline is inserted at Eq. (23), where an exponential wage decay is simply assumed. Substituting that assumed wh into the definition Ph = whLh/(whLh+wAGI LAGI) only restates the assumption in a new variable; the 'nonlinear and accelerating shift' in Section 6 is a property of the chosen exponential ansatz, not a result obtained from the production model. No production function, equilibrium condition, or optimization delivers Eq. (23).

  2. self definitional [Section 5.1, Eqs. (25)-(29)]
    "Since labor is normalized we have Lh + LAGI = 1 ... Express Lh as Lh = 1 − LAGI. Substituting into our power equation yields ... Fully Centralized System (LAGI = 1): Ph = e−λ/(e−λ + (1−e−λ)) ≈ 0 (for large λ)."

    With the normalization Lh = 1 − LAGI, any positive AGI wage wAGI makes Ph tend to zero as LAGI approaches 1, because the human share Lh in the numerator vanishes by construction. The claimed collapse of human economic power at full automation is therefore an artifact of the unit labor endowment identity, independent of the exponential wage assumption and of the production model. Additionally, Eq. (29) drops the factor (1−LAGI) compared with Eq. (27), so the limit '≈ 0 (for large λ)' is not even the correct evaluation of Eq. (27), which is exactly zero for any finite λ.

full rationale

No external benchmarks or machine-checked results are involved. The production-function models in Sections 2-4 are presented as the derivation of wage collapse, but those derivations are not circular in the narrow sense: Eq. (8), Eq. (13), and Eq. (20) are claimed to follow from marginal-product formulas, even though under the paper's own equations (e.g., Eq. (10) with beta1 < 1) w1 diverges as L1 approaches 0 rather than approaching zero; that is a correctness defect, not a circularity. The genuine circularity is in Section 5: the exponential decline of human wages is assumed in Eq. (23) ('We model this decline exponentially') and then used to derive the exponential decline of human economic power in Eqs. (27)-(29). The normalization Lh + LAGI = 1 further builds in Lh toward 0 as LAGI approaches 1, so the 'fully centralized' limit is imposed by definition. No independent evidence for the exponential decay constant lambda is offered, and no equilibrium mechanism connects lambda to the production elasticities of the earlier models. Because the paper's quantitative centerpiece (the 'index function' quantifying accelerating loss of human economic agency) reduces to its own assumption, the score is 7. The paper also contains multiple self-citations to Stiefenhofer and Chen, but they are peripheral to the main derivation and do not load-bear.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The model relies on several unestimated parameters and assumptions. The wage-collapse result is effectively placed into the model by the exponential functional form and the full-substitution assumption, while the Cobb-Douglas marginal product calculation actually contradicts the stated wage conclusion.

free parameters (3)
  • lambda (wage decay constant) = not estimated
    Introduced in Eq. 23 to generate exponential decline of human wages; no estimation or independent justification.
  • w_infinity (asymptotic AGI wage level) = not specified, stated as approaching zero
    Introduced in Eq. 24 as the asymptotic wage of AGI labor; the paper says it 'typically approaching zero' without justification.
  • w_0 (initial human wage) = normalized to 1
    Normalization in Eq. 23 to simplify the model; not a substantive parameter.
assumptions (6)
  • domain assumption Cobb-Douglas production technology with constant elasticities adequately represents production with AGI.
    Used throughout Sections 2 to 4 without justification in the AGI context.
  • standard math Wages equal the marginal product of labor in competitive markets.
    Invoked in Eqs. 7, 10, and 16 to relate wages to derivatives.
  • ad hoc to paper Firms fully substitute human labor with AGI because AGI has zero wage.
    Stated in Section 2.2 and used to take L to zero; ignores adjustment costs, complementarities, and institutional constraints.
  • ad hoc to paper Human labor input may be taken to zero while AGI remains productive.
    Used in Eqs. 5, 13, and 20; not consistent with the original Cobb-Douglas form where output also falls to zero.
  • ad hoc to paper Human wages decline exponentially with AGI labor (Eq. 23).
    Assumed in Section 5.1; this is the result the paper is trying to establish, not a derived consequence.
  • ad hoc to paper Total labor supply is normalized to one: L_h + L_AGI = 1.
    Introduced as Eq. 25 to simplify the power calculation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Artificial General Intelligence and the End of Human Employment: The Need to Renegotiate the Social Contract." pith.science (2026). https://pith.science/paper/K6ORBOHC

@misc{pith2026250207050,
  author       = {Pith},
  title        = {Pith review of: Artificial General Intelligence and the End of Human Employment: The Need to Renegotiate the Social Contract},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/K6ORBOHC}},
  note         = {Machine review of arXiv:2502.07050}
}
read the original abstract

The emergence of Artificial General Intelligence (AGI) labor, including AI agents and autonomous systems operating at near-zero marginal cost, reduces the marginal productivity of human labor, ultimately pushing wages toward zero. As AGI labor and capital replace human workers, economic power shifts to capital owners, resulting in extreme wealth concentration, rising inequality, and reduced social mobility. The collapse of human wages causes aggregate demand to deteriorate, creating a paradox where firms produce more using AGI, yet fewer consumers can afford to buy goods. To prevent economic and social instability, new economic structures must emerge, such as Universal Basic Income (UBI), which redistributes AGI-generated wealth, public or cooperative AGI ownership, ensuring broader access to AI-driven profits, and progressive AGI capital taxation, which mitigates inequality and sustains aggregate demand. Addressing these challenges in form of renegotiation the Social Contract is crucial to maintaining economic stability in a post-labor economy.

Figures

Figures reproduced from arXiv: 2502.07050 by the authors.

Figure 1
Figure 1. Decline of Human Economic Power with Increasing AGI Labor [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

15 extracted references · 13 canonical work pages

  1. [1]

    (1959).Theory of Value: An Axiomatic Analysis of Economic Equilibrium

    Debreu, G. (1959).Theory of Value: An Axiomatic Analysis of Economic Equilibrium . Wiley

    work page 1959

  2. [3]

    Cobb, C., & Douglas, P. (1928). A theory of production.American Economic Review , 18, 139-250

    work page 1928

  3. [4]

    Wicksteed, P. H. (1894).An Essay on the Co-ordination of the Laws of Distribution . 1932 edition, Reprint No. 12. London: London School of Economics

    work page 1932

  4. [5]

    Vale, R. (2024). A note on the Cobb-Douglas function. arXiv preprint . https://doi.org/10.48550/arxiv.2411.08067

    work page Pith review arXiv doi:10.48550/arxiv.2411.08067 2024

  5. [6]

    Oglih, V. (2024). Using the Cobb-Douglas function to analyze economic dynamics.Economic Scope. https://doi.org/10.32782/2224-6282/191-69

    work page doi:10.32782/2224-6282/191-69 2024

  6. [7]

    In Micro-and Nanotechnology Sensors, Systems, and Applications XI (Vol

    Terrile,R.J.(2019).Riseofthemachines: how,whenandconsequencesofartificialgeneralintelligence. In Micro-and Nanotechnology Sensors, Systems, and Applications XI (Vol. 10982). SPIE

    work page 2019

  7. [8]

    Feng, T., et al. (2024). How far are we from AGI.arXiv preprint arXiv:2405.10313

    arXiv 2024

  8. [9]

    (1762).The Social Contract

    Rousseau, J.-J. (1762).The Social Contract . Londres (1964)

    work page 1964

Show all 15 references
  1. [10]

    Applied Mathematical Sciences

    Stiefenhofer, P., &Chen, Y.(n.d.).Industrialartificialintelligence: stabilityofCobb-Douglasproduction functions. Applied Mathematical Sciences

  2. [11]

    Stiefenhofer, P. (2013). The catastrophe map of a two-period production model with uncertainty. Applied Mathematics, 4, 114-121

    2013

  3. [12]

    Stiefenhofer, P. (2014). Topological properties of the catastrophe map of a general equilibrium pro- duction model with uncertain states of nature.Applied Mathematics, 5, 2719-2727

    2014

  4. [13]

    Stiefenhofer, P. (2017). Existence of financial equilibria in a general equilibrium model with piece-wise smooth production manifolds.Journal of Mathematical Finance , 7, 671-681

    2017

  5. [14]

    Stiefenhofer, P. (2016). Production in general equilibrium with incomplete financial markets.Journal of Mathematical Finance , 6, 293-302

    2016

  6. [15]

    Acemoglu, D., & Restrepo, P. (2017). The Race Between Man and Machine: Implications of Tech- nology for Growth, Factor Shares, and Employment.National Bureau of Economic Research Working Paper No. 24174

    2017

  7. [16]

    Korinek, A., & Suh, H. (2024). Artificial General Intelligence and the Future of Labor Demand.arXiv preprint, arXiv:2403.12107. 8

    2024 arXiv

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.