AI and US-Mexico Relations: The Future(s) of Work

Executive Summary

On Feb. 6, 2025, Rice University’s Baker Institute Center for the U.S. and Mexico (CUSMX) held an inaugural symposium for its new AI Policy and Governance initiative.[1] Focused on the theme “AI and the Future(s) of Work,” the event brought together artificial intelligence experts from the United States, Mexico, and other Latin American countries — including Argentina and Colombia — to discuss the impact of AI and automation on workers, labor markets, and on binational labor relations between the U.S. and Mexico.

Participants shared their perspectives on the social, economic, and political impacts of AI in the two countries and identified a series of critical challenges and opportunities for binational collaboration in AI and automation research, development, and policymaking. Group discussions explored shifts in U.S. AI regulatory policy, ongoing labor markets changes produced by generative AI, and emerging opportunities for leadership in responsible innovation. Three key recommendations for AI researchers and policymakers were proposed:

1. New Mechanisms for Policy Innovation and Experimentation — Facilitate cooperation between U.S. and Mexican researchers, policymakers, and civil society through shared data infrastructures, exchange of policy best practices, and the development of regulatory sandboxes and policy prototypes to test labor policies and assess the impact of AI on workers.    
    
2. Human-Centered AI Research — Support collaborative research on the impact of generative AI on labor markets in the U.S. and Latin America, and advance cross-regional efforts to prioritize the augmentation of uniquely human capabilities — such as empathy, creativity, and critical thinking — in AI education and worker retraining initiatives.    
    
3. Neutral Convening Spaces for Ongoing Dialogue — Leverage CUSMX’s AI Policy and Governance initiative as a platform for binational and regional coordination, fostering sustained dialogue and mutually-beneficial policy exchange on AI and the futures of work.

Introduction

Following the public release of ChatGPT in November 2022, a major debate emerged in the United States over the implications of generative AI for the future of work. Some commentators have voiced concerns about generative AI’s impact on cognitive and creative workers, cautioning their potential replacement.[2] Others, however, have expressed optimism, suggesting that the technology’s productive capabilities could enhance cognitive and creative output, resulting in increased wealth and resources.[3]

Although this debate has influenced both U.S. public opinion and policymaking, less attention has been paid to the impact of AI technology on workers outside the U.S. or on how these technologies are currently impacting trade and labor relations between nations.[4] In Latin America, for example, research has revealed different levels of technological adoption compared to the U.S. across the two previous waves of AI-driven transformation: lower adoption of industrial robots during the first wave and a higher uptake of on-demand labor platforms in the second.[5]

Now, with the onset of a third wave of AI-driven transformation fueled by generative AI, these differences raise new and important questions, such as:

• Whose future is being considered in conversations about AI and the future of work?
• How are multiple different futures — rather than a single trajectory — being shaped by AI technologies?
• How are global labor markets and societies being affected by changes to the future of work in the U.S., and how are these developments influencing the U.S. in return?

Symposium

With these questions in mind, on Feb. 6, 2025, the Center for the U.S. and Mexico (CUSMX) at Rice University’s Baker Institute for Public Policy held a symposium — the first in a series of events on AI and U.S.–Mexico relations — focused on the theme “The Future(s) of Work.” The symposium had two main purposes:

1. To launch the center’s new initiative of research on AI policy and governance. This initiative is led by this report’s authors, with the support of Tony Payan, CUSMX director.    
    
2. To create a space for dialogue on the questions above with leading AI experts from the United States, Mexico, and Latin America.

To advance these goals, CUSMX’s AI policy and governance team convened a group of 18 scholars, policymakers, and leaders from civil society, and industry — including representatives from Brown University, MIT, PIT Policy Lab, Centro-I para la Sociedad del Futuro, Eon Institute, the Global Network of Internet and Society Research Centers, Bitso, Microsoft, Innit, the Inter-American Development Bank, CAF (Development Bank of Latin America and the Caribbean), the state of Jalisco (Mexico), and the Consulate General of Mexico in Houston.

During the symposium, participants explored issues through three focused dialogues:

1. U.S. and Mexico Perspectives on AI gathered broad insights on recent AI developments in Mexico and the United States, focusing on their impact on socioeconomic trends.   
    
2. Future(s) of Work examined how AI is reshaping the nature of work and affecting labor markets and workers in both Mexico and the U.S.   
    
3. Responsible AI Innovation and Leadership explored the role of civic and corporate leaders in shaping responsible AI development and advancing ethical governance frameworks.

This report summarizes the three symposium dialogues and concludes with key takeaways and recommendations for AI researchers and policymakers in the region. All information is presented in accordance with the Chatham House Rule, under which neither the identity of speakers nor their affiliations are disclosed.

Dialogue I: US and Mexico Perspectives on AI

The first dialogue reflected on the current state of AI readiness in the United States and Mexico, raising critical questions about the technology’s potential social, economic, and political impacts in both countries.

Mexican Focus

The discussion began with a focus on Mexico. Despite being the world’s 12th largest economy and a key regional player, Mexico has fallen behind other countries in Latin America in developing and executing a national AI agenda.[6] This gap in AI capabilities and regulation was discussed in relation to three factors currently limiting AI readiness in Mexico: 

1. Public Infrastructure Access — Broader access to data and computing infrastructure is essential for researchers and state-level representatives seeking to address local needs.    
    
2. Opportunities for Professional Development — While Mexico ranks seventh globally in STEM graduates, it faces significant human capital flight, as many skilled professionals emigrate due to regional economic insecurity.[7]    
    
3. Ethical Frameworks — Clearer guidance is needed across both the public and private sectors to ensure AI development does not deepen existing social inequalities. 

Despite these challenges, participants also praised recent governance developments in Mexico, including the creation of the Agency for Digital Transformation and Telecommunications (ADTT), along with AI-focused initiatives such as the National AI Alliance, which aims to strengthen the country’s AI ecosystem.[8] In addition, optimism was expressed around growing Mexico’s AI capabilities, given recent efforts to bolster chip design and technological capabilities. DeepSeek was mentioned as an example of how less resource-intensive AI models could be soon become viable for developers in Mexico — provided they have access to sufficient computing power and equipment.[9]

US Focus

With regard to the U.S., the group discussion focused on recent regulatory changes under the second Trump administration, including the revocation of a 2023 Biden administration executive order aimed at promoting secure, ethical, and trustworthy AI practices.[10] While there was general continuity in AI strategy between Donald Trump’s first administration and Joe Biden’s presidency, the shifts in Trump’s second term are substantial.

A comparison of the 2023 and 2025 executive orders on AI revealed notable changes — from language emphasizing “responsibility” to a narrative centered on “global AI dominance,” and from addressing “bias and discrimination” to prioritizing “national security.”[11] In context of recent realignments in American foreign policy and the emergence of trade-related tensions between Mexico and the U.S., participants expressed concerns for the future of international cooperation between the two nations on AI research and development.

However, there was some optimism regarding state-level regulation in the U.S. For example, Colorado and Utah have passed artificial intelligence acts that actively commit resources to AI development, while New Hampshire has passed legislation aimed at addressing specific risks, such as deepfakes.[12]

Binational Approach

In terms of binational relations, the importance of academic and private industry-level cooperation was emphasized. The discussion noted how the U.S. has historically benefited from Mexican labor and the import of material resources to aid technological development. From the Mexican perspective, advancing national AI capabilities could help drive progress across Latin America.

Key challenges identified included the absence of AI strategies, the need for high-capacity computing infrastructure, and limited collaboration between academia and industry. Recent regional initiatives — such as the declarations of Santiago and Montevideo — were highlighted as steps toward deeper collaboration in AI research, development, and policymaking.[13]

Dialogue II: AI and the Future(s) of Work

The next dialogue focused on AI within the context of the future of work, highlighting three stages in AI’s development and impact:

1. Industrial Robots — Deployed world-wide in manufacturing plants to automate repetitive tasks.   
    
2. Digital Platforms — Services like Uber and DoorDash, where technology companies leverage informal labor markets to provide on-demand consumer services.   
    
3. Generative AI — AI capability extending beyond low-skill tasks to medium-skill cognitive functions.

Impact of Generative AI 

The discussion on this third stage underscored the need to accurately assess how generative AI is affecting labor markets in Mexico and Latin America, citing an Inter-American Development Bank (IDB) report that focused on large language models (LLMs) — a widely-used form of generative AI — to evaluate impacts on the labor markets of Chile, Mexico, and Peru.[14] The report found a relatively low level of exposure to generative AI across the three countries, but also described a potentially greater exposure among specific social groups, including:

• Women compared to men.
• Individuals with higher education levels over lower education levels.
• Formal over informal employees.
• Higher-income groups relative to lower-income groups.

The report speculates on the potential positive or negative impacts. For instance, greater exposure to AI could lead to automation and loss of employment, such as in administrative positions, which more frequently held by women. In other cases, it could enhance productivity, particularly in highly-skilled jobs typically associated with higher education and income levels. The participants expressed concern that AI could worsen existing inequities.

The finding that workers in the informal sector had lower AI exposure challenged earlier narratives that AI mainly affects low-skilled jobs and also demonstrated the growing impact of LLMs on formal employment. There was a call for further research to better understand these differences, both in Mexico and other Latin American countries not involved in the original study.

The dialogue turned to recent work by MIT researchers Isabella Loaiza and Roberto Rigobon on “augmentation” — using AI to enhance human capabilities.[15] Their research reframes traditional measures of AI exposure into a concise framework that captures both risks and benefits, using the EPOCH model, which focuses on AI’s potential to strengthen what they describe as “human intensive capabilities,” including:

• E — Empathy and emotional intelligence.
• P — Presence and human contact.
• O — Opinion and judgement.
• C — Creativity and imagination.
• H — Hope and vision.

This research demonstrated a positive correlation between the usage of AI and EPOCH measures for U.S. workers in the period 2016–24, but the framework has not yet been applied to workers outside the U.S. Mapping potential augmentations in these categories to relevant jobs in Latin American countries could help better understand the impacts of automation and augmentation on the region’s workforce.

 National AI Policy

 There are ongoing tensions around AI regulation and innovation in both countries. In the U.S., recent AI policy changes give rise to questions such as:

• Who will lead AI development — academia or industry — and how will that shape future hiring?
• How will AI influence education, including which skills young people will pursue and which criteria employers will use in hiring?
• Will responsible AI development guidelines be established through centralized (federal) or decentralized (state-level or private) initiatives? 

In Mexico, a recent reform passed by the senate aims to guarantee on-demand workers social security benefits similar to those of full-time employees.[16] It remains unclear whether this law will protect workers or, by raising consumer prices, reduce job opportunities for low-income workers.

Dialogue III: Responsible AI Innovation and Leadership

The final dialogue explored current possibilities for responsible AI innovation and leadership in responsible AI policy work.

The work developed by the Global Network of Centers (NoC) is a positive example for the role of international cooperation in responsible AI policymaking. In recent years, the NoC has worked closely with academic institutions, governments, and civil society organizations to design frameworks that reflect local needs and capabilities, while aligning with global best practices.[17] As part of these efforts, the NoC has helped to create AI regulatory sandboxes in Brazil and Chile, organize the UN Summit of the Future (2024), and develop the first-ever Latin America AI Index (2024).[18]

Globally, the NoC has also focused on supporting the development of public policies, including the design of national AI strategies that emphasize talent development by promoting inclusive education and training programs, reducing skills gaps, and creating pathways for researchers and professionals from diverse backgrounds to actively contribute to the AI ecosystem.

The discussion also underscored the importance of bringing Latin American perspectives into international AI dialogue. The relative absence of major headlines on AI-related issues in the region — such as bias, discrimination, hallucinations, and technological unemployment — in global news was noted. Rather than indicating a lack of such incidents, this raised concerns about underreporting and highlighted the need for greater visibility and understanding of how AI can affect Latin America.

A three-pronged approach was proposed to enhance awareness of AI-related risks in Mexico and other Latin American countries, including recommendations for researchers to:

• Map known AI-related incidents in crowd-based sources, such as the AI Incident Database.
• Share best practices and raise questions through communications, such as the AI Action Summit’s International AI Safety Report.
• Develop new AI strategies tailored to each country’s current capabilities.

The importance of Mexico formulating an official national AI strategy was emphasized, with greater focus on developing semiconductor production capacity and other key resources to support international cooperation and development.

Examples of how corporations can lead in promoting responsible AI usage and developing responsible AI guidelines include: 1) Microsoft’s recent Accelerated Foundation Models Research (AFMR) initiative, which has raised opportunities for minority groups in the United States; 2) Innit’s development of health-based AI-powered software that provides personalized recommendations and uses expert nutritionist advice to fact-check generative AI results; 3) Bitso’s use of blockchain-based systems for U.S.-Mexico remittances, which saw a 90% increase in transaction volume over its first year and now processes 10% of all remittances between the two countries.[19]

Conclusion and Recommendations

Symposium dialogues explored recent developments in responsible AI regulation, the impact of AI on Latin American workers, and the private sector’s role in advancing responsible AI innovation. Drawing from these discussions, this report offers three key recommendations for AI researchers, policymakers in Mexico and the U.S., and other stakeholders:

 1. Develop New Mechanisms for Policy Innovation and Experimentation

Given ongoing U.S. policy movement and uncertainty about the impact of AI on U.S. and Mexican workers, it is important to cultivate continuing collaboration among the global community of AI scholars and policymakers — to share data, exchange best practices, and seek jointly beneficial mechanisms to improve both human and technical capabilities.

The first recommendation encourages the development of experimentation spaces — such as sandboxes or policy prototypes — in both the U.S. and Mexico.[20] These environments allow for the safe testing and assessment of potential regulatory measures, enabling concrete analysis of realistic use-cases and their impacts before scaling to broader populations.

The value of experimental spaces for AI and automation is illustrated by the Institute for the Future of Work (IFOW) which has helped pilot regulatory sandboxes to advance workers’ rights in the U.K. Ongoing collaboration between IFOW and the Global Network of Centers will provide an opportunity to leverage this expertise to better understand AI’s impacts on U.S. and Mexican workers and to safely test potential regulatory approaches in both countries.[21]

2. Promote Human-Centered AI Research on the Futures of Work

Loaiza and Rigobon’s EPOCH framework identifies key human-intensive capabilities — empathy, presence, opinion, creativity, and hope — that complement AI’s current limitations.[22] Their research offers a path for policymakers and industry leaders in Mexico and the U.S. to prioritize initiatives that enhance these abilities and support more effective AI education and workforce training.

To gain a deeper understanding of the potential impacts of AI on American and Mexican workers, further research is recommended on:

• The impacts of generative AI on medium- and high-skilled workers across the Americas.
• The augmentation of human labor, in addition to the effects of automation.

This research should expand on the IDB’s research in Mexico, Peru, and Chile by including comparative analysis across the U.S., Mexico, and other Latin American countries.[23] It should generate robust, context-specific evidence on how emerging AI technologies — particularly large language models — are reshaping labor markets across the Americas, highlighting opportunities for continued collaboration in research, capacity-building, and policymaking.

3. Create Neutral Convening Spaces for Ongoing Dialogue

Building on the previous proposals, the third recommendation entails supporting the establishment of neutral, research-based convening spaces that foster binational and regional dialogue and promote evidence-informed policymaking.

The newly established AI Policy and Governance initiative at Rice University’s Baker Institute for Public Policy was identified as a strong platform for advancing U.S.-Mexico collaboration, particularly by bridging research and practice. As a trusted academic institution with deep regional ties, the Baker Institute is well positioned to host policy roundtables, multidisciplinary research exchanges, and stakeholder consultations that bring together experts from government, industry, media, academia, and civil society. 

These convenings would be key to building trust, facilitating data sharing, and translating global and regional research — such as that from IFOW and the EPOCH framework — into context-specific guidance for policymakers and practitioners. Institutionalizing this role would help ensure continuity and encourage AI policy development across the U.S., Mexico, and beyond.

Acknowledgements

The authors thank Alejandro Dabdoub for his generous support of the symposium and the AI Policy and Governance initiative, Michelle Moya Aceves for her assistance in preparing this report, and all symposium participants for their valuable contributions.

Notes

[1] “AI Policy and Governance,” Center for the U.S. and Mexico, Baker Institute for Public Policy, accessed May 21, 2025, https://www.bakerinstitute.org/ai-policy-governance.

[2] Ozge Demirci et al., “Research: How Gen AI Is Already Impacting the Labor Market,” Harvard Business Review, November 11, 2024, https://hbr.org/2024/11/research-how-gen-ai-is-already-impacting-the-labor-market; Matt Egan, “AI Is Replacing Human Tasks Faster than You Think,” CNN Business, June 20, 2024, https://www.cnn.com/2024/06/20/business/ai-jobs-workers-replacing/index.html; and Pranshu Verma, “ChatGPT Provided Better Customer Service than His Staff. He Fired Them,” Washington Post, October 3, 2023, https://www.washingtonpost.com/technology/2023/10/03/ai-customer-service-jobs/.

[3] Martin Neil Baily et al., “Machines of Mind: How Generative AI Will Power the Coming Productivity Boom,” Brookings, May 5, 2023, https://www.brookings.edu/?p=1687743&post_type=article&preview_id=1687743; Piyachart Isarabhakdee, “AI Will Drive Growth. But Only Authentic Intelligence Can Empower the World,” World Economic Forum, March 20, 2025, https://www.weforum.org/stories/2025/03/ai-authentic-intelligence; and McKinsey & Company, “The Economic Potential of Generative AI: The Next Productivity Frontier,” June 14, 2023, https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/the-economic-potential-of-generative-ai-the-next-productivity-frontier.

[4] Notable exceptions include work by researchers and journalists at Rest of World, as well as organizations like Data & Society and the Carnegie Endowment for International Peace. For examples, see: Sareeta Amrute et al., “A Primer on AI In/from the Majority World,” Data & Society, December 14, 2022, https://datasociety.net/library/a-primer-on-ai-in-from-the-majority-world/; “Artificial Intelligence and the Global Majority,” Carnegie Endowment for International Peace, accessed April 19, 2025, https://carnegieendowment.org/programs/technology-and-international-affairs/artificial-intelligence-and-the-global-majority?lang=en; and “Mexico Is Using an AI-powered App to Prevent Suicides,” Rest of World, December 12, 2024, https://restofworld.org/series/the-rise-of-ai/.

[5] Laura Ripani et al., The Future of Work in Latin America and the Caribbean: What is the Impact of Automation on Employment and Wages? (2020), https://doi.org/10.18235/0002960; Economic Commission for Latin America and the Caribbean (ECLAC) and International Labour Organization (ILO), “Decent Work for Platform Workers in Latin America,” Employment Situation in Latin America and the Caribbean, no. 24 (June 2021), https://repositorio.cepal.org/server/api/core/bitstreams/5484c1bf-c259-4a9a-aa98-22db13d52a8b/content.

[6] “Mexico,” The Observatory of Economic Complexity, accessed April 18, 2025, https://oec.world/en/profile/country/mex; “LATAM Artificial Intelligence Index 2024,” Indice Latam, accessed April 18, 2025, https://indicelatam.cl/home-en-2024/.

[7] Secretariat of Economy et al., Mexican Talent for Economic Growth and Nearshoring, April 2023, https://www.gob.mx/cms/uploads/attachment/file/828153/talento-mexicano_ing.pdf; Oxford Analytica, “Mexico’s Brain Drain to Continue Unabated,” Expert Briefings, May 18, 2018, https://doi.org/10.1108/OXAN-DB233864.

[8] Alianza Nacional de Inteligencia Artificial (ANIA), homepage, accessed April 18, 2025, https://www.ania.org.mx/.

[9] Cade Metz, “What to Know About DeepSeek and How It Is Upending AI,” New York Times, January 27, 2025, https://www.nytimes.com/2025/01/27/technology/what-is-deepseek-china-ai.html.

[10] The White House, “Initial Rescissions of Harmful Executive Orders and Actions,” January 20, 2025, https://www.whitehouse.gov/presidential-actions/2025/01/initial-rescissions-of-harmful-executive-orders-and-actions/.

[11] Joseph R. Biden, “Executive Order 14110, Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence,” Federal Register, October 30, 2023, https://www.federalregister.gov/documents/2023/11/01/2023-24283/safe-secure-and-trustworthy-development-and-use-of-artificial-intelligence; Donald J. Trump, “Removing Barriers to American Leadership in Artificial Intelligence,” White House, January 23, 2025, https://www.whitehouse.gov/presidential-actions/2025/01/removing-barriers-to-american-leadership-in-artificial-intelligence/.

[12] S.B. 24-205, 65th Leg., First Reg. Sess. (Colo. 2024), https://leg.colorado.gov/bills/sb24-205; S.B. 149, 65th Leg., Gen. Sess. (Utah 2024), https://le.utah.gov/~2024/bills/static/SB0149.html; and H.B. 1432, 169th Leg., Reg. Sess. (N.H. 2024), https://legiscan.com/NH/text/HB1432/2024.

[13] Santiago Declaration, November 8, 2024, https://conferenciaelac.cepal.org/9/sites/elac9/files/2401157e_cmsi.9_santiago_declaration.pdf; Montevideo Declaration, October 2024, https://www.gub.uy/agencia-gobierno-electronico-sociedad-informacion-conocimiento/sites/agencia-gobierno-electronico-sociedad-informacion-conocimiento/files/documentos/noticias/EN%20-%20Montevideo%20Declaration%20approved.pdf.

[14] Orazio Azuara Herrera et al., AI and the Increase of Productivity and Labor Inequality in Latin America: Potential Impact of Large Language Models on Latin American Workforce (2024), https://doi.org/10.18235/0013152.

[15] Isabella Loaiza and Roberto Rigobon, The EPOCH of AI: Human-Machine Complementarities at Work, MIT Sloan Research Paper No. 7236-24, November 21, 2024, https://ssrn.com/abstract=5028371.

[16] Fernanda Gonzalez, “Senadores aprueban prestaciones de ley a conductores de apps como Uber y Didi” [Senators Approve Reform That Provides Legal Benefits to App Drivers Like Uber and Didi], Wired España, December 13, 2024, https://es.wired.com/articulos/senadores-aprueban-prestaciones-de-ley-a-conductores-de-apps-como-uber-y-didi.

[17] Global Network of Internet and Society Research Centers (NoC), NoC 2024 Progress Report, https://networkofcenters.net/sites/networkofcenters.net/files/NoC%202024%20Progress%20Report%20.pdf.

[18] This helpful definition is from the government of Canada (“Regulatory Sandbox,” Modernizing Regulations, government of Canada, modified July 15, 2024, https://www.canada.ca/en/government/system/laws/developing-improving-federal-regulations/modernizing-regulations/regulatory-sandbox.html): “A regulatory sandbox is a tool that can help federal regulators keep pace with changing technologies and reflect current business realities, challenges, and opportunities. Regulatory sandboxes can allow industry to demonstrate the real-life impacts of a new product or service in the marketplace under a temporary set of rules and controlled by regulatory supervision. This can help a regulator safely decide whether to make any permanent changes to how that product or service should be regulated.”

[19] “Announcing Recipients of the AFMR Minority Serving Institution Grant,” Microsoft Research Blog, January 30, 2024, https://www.microsoft.com/en-us/research/blog/announcing-recipients-of-the-afmr-minority-serving-institutions-grant/?msockid=079cc708636d66070e62d3e8623a6704; Innit, “Innit Unveils Validators for Personalized Food Intelligence, Available on Google Cloud Marketplace,” Accesswire, news release, April 23, 2024, https://www.innit.com/newsDetail?data=ZjC8_BIAADUe4sVf; Finextra, “Bitso Surpasses $12 Billion in Transactions in 2024,” December 17, 2024, https://www.finextra.com/pressarticle/103639/bitso-surpasses-12-billion-in-transactions-in-2024.

[20] Armando Guio, Regulatory Sandboxes in Developing Economies an Innovative Governance Approach, Economic Commission for Latin America and the Caribbean (ECLAC), 2024, https://repositorio.cepal.org/server/api/core/bitstreams/960a7e7d-ab60-4c3e-b4b2-c7a302e2ea16/content.

[21] “Responsible AI Sandbox,” Institute for the Future of Work, https://www.ifow.org/landing-page/sandbox.

[22] Loaiza and Rigobon.

[23] Azuara Herrera et al.

This publication was produced on behalf of Rice University’s Baker Institute for Public Policy. Wherever feasible, the material was reviewed by external experts prior to its release. Any errors are the responsibility of the author(s) alone.

This material may be quoted or reproduced without prior permission, provided appropriate credit is given to the author(s) and Rice University’s Baker Institute for Public Policy. The views expressed herein are those of the individual author(s) and do not necessarily represent the views of Rice University’s Baker Institute for Public Policy.