Who says what
Researchers’ documented positions on “Will AI widen inequality?”
Documented positions only, coded with cited evidence by an AI agent: 27 of 400 mapped researchers so far. Everyone else is unknown: no reviewed evidence, which is neither agreement nor neutrality.
Agree (23)
Anton Korinek agrees
A core theme of his research: without policy, advanced AI concentrates income and worsens distribution, in advanced and developing economies.
Benjamin Moll agrees
Co-authored theory and evidence that automation increases income and wealth inequality by raising returns to wealth, so its benefits accrue disproportionately to owners of capital.
Christopher A. Pissarides agrees
The review he led on automation and work found technological change compounding existing inequalities and creating new ones across groups and places, and called for systematic policy action.
Evidence
- “Left unchecked, technological transformation will further entrench regional inequalities.” — source
Daniel Trefler agrees
His 2026 policy paper argues that, as with earlier scalable technologies, dominant AI firms are positioned to capture large profits and political influence that aggravate social inequality, which he presents as a case for regulation.
Daron Acemoglu agrees
A central finding of his research: automation has been a major driver of wage inequality, and generative AI's benefits will likely be distributed unevenly without policy.
David Hémous agrees
His growth models and patent evidence link automation innovation to rising income inequality.
Joseph Stiglitz agrees
His AI papers identify inequality as a main challenge of worker-replacing technological progress and set out domestic and international policies to mitigate the adverse distributional effects.
Klaus Prettner agrees
His macro models and empirical work find automation raises inequality and depresses the labor share without policy intervention.
Pascual Restrepo agrees
His work attributes a large share of rising US wage inequality to automation displacing routine task workers.
Anna Salomons leans agree
Documents automation eroding the labor share of income across countries.
Arvind Karunakaran leans agree
His co-authored 2025 review of research on AI and workplace inequality proposes an 'inequality cascade' in which choices in AI design, implementation and use can amplify encoded, evaluative, wage and relational inequality, and discusses mechanisms to mitigate it.
Evidence
- Artificial Intelligence at Work: An Integrative Perspective on the Impact of AI on Workplace Inequality
- “Choices made during model design, implementation, or use and adaptation can ripple out to exacerbate or reduce workplace inequality.” — source
Bryan Seegmiller leans agree
Co-authored studies document that technological change has raised demand for higher-education, higher-wage occupations and that new work since 1980 has concentrated in high-paid professional roles; these are historical patterns rather than an AI-specific forecast.
Christian Catalini leans agree
The AGI paper argues that the gains of AI deployment are privatized while systemic risks are socialized, and that rents shift toward those who verify and underwrite AI output.
Evidence
- “the gains of AI deployment are aggressively privatized, while the systemic risks are socialized” — source
Christos Makridis leans agree
Finds AI exposure lowers the labor share, with workers capturing only part of AI-driven output growth, and examines broad employee ownership as a way for workers to share in AI-related surplus.
Evidence
- Who owns the gains from AI? Employee ownership, tokenization, and the distribution of income
- “Workers are capturing only about 29 cents of every dollar of AI-driven output growth” — source
Daniel Susskind leans agree
Argues that if AI reduces demand for human work, a central problem becomes distributing prosperity when wages are a less effective channel, and his book proposes a larger redistributive role for the state.
Evidence
- “There is the problem of distribution – how to share prosperity in society when our traditional way” — source
- A World Without Work: Technology, Automation and How We Should Respond
Diane Coyle leans agree
Says the value from AI is currently captured by a small number of people and that political disruption will follow unless that value is shared more widely.
Evidence
- “Unless that value starts being shared, we're going to be in very politically disrupted times.” — source
Georgios Petropoulos leans agree
He finds AI talent and benefits concentrated in a small number of large firms and argues that market regulation, competition and labour-market policies are needed; his work also links automation to fewer jobs for less-educated workers.
Evidence
- “This leads to only a small portion of firms capturing most of the talent and the benefits from AI technologies” — source
- Automation, Covid-19, and Labor Markets
James Landay leans agree
He has said the gains and losses from AI-driven job change will not be distributed evenly and calls for a human-centered approach to these societal effects.
Evidence
- “the gains and losses are not going to be distributed evenly” — source
Maria del Rio-Chanona leans agree
Freelance-platform evidence finds losses concentrated in substitutable skills and among novice workers.
Martin Beraja leans agree
His 'inefficient automation' work implies laissez-faire AI adoption skews gains away from workers.
Evidence
Neil Thompson leans agree
Has said an increasing share of income is likely to flow to capital owners as AI spreads, which he describes as a societal challenge; he does not specify a policy response in the source found.
Evidence
- “an increasing share of income is likely to flow to capital owners” — source
Rembrand Koning leans agree
Finds generative AI's benefits are uneven — across entrepreneurs and along global gender gaps in adoption.
Simon Johnson leans agree
Co-authored work arguing that automation raises wages only alongside new tasks or complementary hiring and that wages are unlikely to rise when workers cannot claim a share of productivity growth; a co-authored review states AI's workplace benefits will likely be distributed unevenly.
It's complicated (3)
Ajay Agrawal it's complicated
Suggests augmenting AI could compress skill premiums by leveling expertise — but notes automation-heavy paths do the opposite.
David Autor it's complicated
Documented technology-driven polarization for decades, yet argues AI could rebuild middle-skill work by extending expertise — direction depends on how AI is deployed.
David J. Deming it's complicated
An experiment finds AI narrows performance gaps between workers, while survey work documents large cross-country gaps in AI adoption.
Disagree (1)
Ben Weidmann leans disagree
Co-authored experiment finds AI assistance improves performance more for people with low baseline ability and reduces performance inequality. This concerns task performance in an experiment, not income or wealth inequality.