Mary L. Gray

Mary L. Gray is Senior Principal Researcher at Microsoft Research (since 2012), Faculty at Indiana University's Luddy School of Informatics, Computing, and Engineering (with affiliations in Anthropology, Gender Studies, and American Studies), and Faculty Associate at Harvard's Berkman Klein Center for Internet and Society (see Person entry). She is the anthropologist-academic public voice most consistently credited with naming and documenting the AI supply chain's structurally concealed human-labor remainder — through Ghost Work: How to Stop Silicon Valley from Building a New Global Underclass (Houghton Mifflin Harcourt, 2019, co-authored with Siddharth Suri), which coined both "ghost work" and "automation's last mile" as the foundational terms for this labor class; through the 2020 MacArthur Fellowship that anchored her as the named scholar of digital-economy labor transformation; and through continuing policy-advisory roles — chairing Microsoft Research's Ethics Review Program and serving on the California Governor's Council of Economic Advisors and Stanford's AI100 Standing Committee — that translate the Ghost Work argument into AI-governance channels. She is tracked here as a Voice because her named public output — the Ghost Work book and the "ghost work" framing it installed across global worker-organizing registers, the ongoing ethics-review and policy-advisory practice, the speaking circuit carrying the platform-labor argument into AI-governance venues — is the load-bearing object the corpus needs to track alongside the publication-side anchor and message entry already in the corpus.

Voice anchor

Gray's Voice closes the AI supply chain / platform labor ethnographer-academic public-voice slot. The corpus had pub-ghost-work-gray-suri-2019 as the publication-side anchor for the ghost-work framing and msg-ghost-work as the message tracking the framing's propagation through global worker organizing, but no Voice carrying the ongoing public-output practice — the ethics-review and policy-advisory channels, the speaking circuit, the continuing anthropological research at Microsoft — that has extended that framing from a 2019 book into a continuing AI-governance posture. Three distinctions from adjacent voices already in the corpus:

• Distinct from Virginia Eubanks's Voice. Eubanks anchors the organizer-academic register of the US welfare-state algorithmic-harm layer — a researcher who spent two decades simultaneously doing grassroots welfare-rights organizing in the same communities she studies, whose authority rests on that dual ground-level presence. Gray anchors the researcher-academic register of the AI supply chain labor layer — a Microsoft Research anthropologist whose authority rests on the five-year ethnographic base of the Ghost Work study, the institutional reach of the MSR Ethics Review Program, and policy-advisory access through the Governor's Council and AI100. The structural positions are complementary: Eubanks works from inside affected communities outward; Gray works from inside a major AI company outward. Both name a layer of algorithmic harm to workers; neither names the same layer or occupies the same institutional position.

• Distinct from Timnit Gebru's Voice. Gebru anchors the independent-African-diaspora-AI-research register — the DAIR model of AI research outside Big Tech's incentive structures, with the 2020 Google dismissal as the structural anchor. Gray's register is embedded within Big Tech (Microsoft Research since 2012), carrying the critique of platform-labor conditions from inside one of the companies Ghost Work named. The structural positions are opposite: Gebru's Voice is defined partly by institutional rupture with Big Tech; Gray's Voice operates from within Big Tech while maintaining the external critique and policy-advisory practice simultaneously. Different sites of authority; different organizational models.

• Distinct from Safiya Noble's Voice. Noble anchors the library-and-information-science × algorithmic-oppression register — how search engines and information-organisation infrastructure encode racial and gender hierarchy, from the LIS and Black-feminist traditions. Gray anchors the AI supply chain × platform labor register: the structural concealment of the human workforce that AI systems require, and the labor conditions that workforce experiences. Noble's subject is the knowledge infrastructure; Gray's subject is the production infrastructure. Different layers of the AI system; different analytical traditions (LIS vs. cultural anthropology and STS); different movement organising registers.

Signature framings

Two framings anchor Gray's public-output register and have carried the most weight in the AI supply chain and make-AI-good movement vocabulary.

• "Ghost work." The central term of Ghost Work — coined by Gray and Siddharth Suri for task-based human labor that AI systems and digital platforms structurally require but systematically conceal, routed through APIs and algorithmic interfaces that render the worker invisible to the end user while presenting the transaction as human–machine interaction. As Gray formulated it for MIT Technology Review in May 2019, ghost work is "any work that could be — at least in part — sourced, scheduled, managed, shipped, and built through an application programming interface" — and it becomes ghost work specifically when companies "market the output as fully automated while relying on hidden human labor." The political charge: work is not disappearing in the age of AI; it is being hidden. The invisibility is not incidental to the platform design; it is the most load-bearing feature of the platform's commercial identity — the product being sold is "AI," not "AI plus a large underpaid workforce." Fully documented in the msg-ghost-work message entry; tracked here as Gray's foundational framing contribution to the movement vocabulary.

• "Automation's last mile." The companion analytical concept from Ghost Work — naming the irreducible remainder of tasks that AI systems cannot yet fully automate, offloaded onto a distributed human workforce through API interfaces. The "last mile" metaphor adapts the logistics-infrastructure concept: as the last mile of package delivery requires physical, judgment-intensive human labor that algorithmic routing cannot replace, the last mile of AI automation requires human judgment for edge cases, cultural-context calls, and error-correction that neural networks cannot reliably perform. The MacArthur Foundation's 2020 Fellowship page describes Gray's research as documenting "the paradox of automation's last mile" — a paradox because the more AI systems advance, the more they generate last-mile remainder tasks that require the human labor they are nominally replacing. The framing is the structural explanation for why ghost work is not a transitional phenomenon en route to full automation but a permanent architectural feature of any AI production system.

Public output and venues

Gray's named public-output channels run through four overlapping registers.

• The Ghost Work book. Ghost Work: How to Stop Silicon Valley from Building a New Global Underclass (Houghton Mifflin Harcourt, 2019) is the primary corpus anchor — winner of the 2019 McGannon Center Book Prize at Fordham University, named a Financial Times 2019 Critic's Pick, and translated into Korean and Chinese — placing it in the international circulation of AI-labor discourse beyond English-language academic and policy venues. The ten policy and technical recommendations the book closes with — portable benefits, data portability, minimum-wage and overtime protections for task-based work, platform-design requirements for worker visibility and collective-representation capacity — are the programmatic form of the ghost work framing that subsequent organizing vehicles (the Data Labellers Association, African Content Moderators Union, Alphabet Workers Union–CWA) have drawn against.

• Institutional ethics-governance channel. Gray chairs Microsoft Research's Ethics Review Program — the internal review structure for Microsoft's research agenda on AI and society. This is the primary channel through which her anthropological analysis of platform-labor conditions feeds into the AI-governance architecture of one of the companies Ghost Work identified as structurally reliant on ghost work. The structural position is deliberate and institutionally unusual: the researcher who documented ghost labor conditions in platforms including Microsoft-adjacent ones is also the person who chairs the review program through which Microsoft assesses the social implications of its research. The same scholar occupies both the external critique register (Ghost Work, the MacArthur Fellowship) and the internal governance register (Ethics Review Program chair) — a dual-position that gives her policy reach that a purely external critic would not have.

• Policy-advisory roles. Gray's named advisory positions include the California Governor's Council of Economic Advisors, Stanford's AI100 Standing Committee, and PRIM&R (Public Responsibility in Medicine and Research). These translate the Ghost Work argument into state-level labor-economics policy (Governor's Council), long-horizon AI-governance research (AI100 — the Stanford study tracking AI's progress and societal impact over a hundred years), and research-ethics governance (PRIM&R) — three channels through which the platform-labor argument reaches policymakers and governance infrastructures beyond the academic and public-advocacy circuits.

• The earlier ethnographic register. Gray's prior book — Out in the Country: Youth, Media, and Queer Visibility in Rural America (MIT Press, 2009), winner of the American Anthropological Association's Ruth Benedict Prize and the American Sociological Association's Sexualities Studies Book Award — established the methodological register that Ghost Work extends. The earlier book's argument: that LGBTQ rural youth were using digital media as an identity-formation and community-building infrastructure in the absence of visible local queer community, rendered simultaneously networked and invisible within platforms that recorded their presence without acknowledging it. The methodological continuity with Ghost Work is direct: both books document populations rendered structurally invisible by digital infrastructures whose value depends on concealing the human presence that sustains them. The invisibility thesis runs from rural queer youth (2009) through platform ghost workers (2019) as the same anthropological question applied to different populations and digital contexts.

Why this is a Voice entry

A Voice entry is created here, rather than additional structure on the Person entry, because Gray's named public output is the load-bearing object the corpus needs to track: Ghost Work and the "ghost work" framing it installed in AI-accountability and make-AI-good movement vocabulary; the msg-ghost-work framing's global propagation through African and US worker organizing registers; the 2020 MacArthur Fellowship anchoring her as the named scholar of digital-economy labor transformation; the MSR Ethics Review Program chairship translating the research into the internal governance structure of a major AI company; and the Governor's Council and AI100 service translating it into state-level labor-economics and AI-governance channels. The corpus had pub-ghost-work-gray-suri-2019 as the publication-side anchor for the ghost-work framing and msg-ghost-work as the message tracking the framing's organizational propagation — but no Voice carrying the ongoing public-output practice that has extended those corpus objects into continuing policy channels and governance institutions. This entry closes the asymmetry. The distinctive register the Voice closes — anthropologist-academic embedded within Big Tech while maintaining an external critique and policy-advisory practice simultaneously — has no other carrier in the corpus's current Voice coverage. Affiliation, training, and biographical detail are recorded on the linked Person entry per the corpus's Person/Voice split.