Summer Sullivan (UC Santa Cruz) — SSRC Just Tech essay on Salinas Valley romaine lettuce harvest automation; substantive US critical voice on specialty-crop AI and farmworker labour

NA-US (Salinas Valley California; Monterey County)

Content

Summer Sullivan is a PhD candidate in Sociology at the University of California, Santa Cruz. Her September 10, 2025 essay “A Tale of Two Valleys: The Salad Bowl of the World Goes Digital”, published in the Social Science Research Council’s Just Tech publication (DOI 10.35650/JT.3087.d.2025), is the most substantive recent critical-voice contribution on US specialty-crop AI and farmworker labour in the agrifood AI field. Per memory hygiene (substantive analytical contribution over citation volume): Sullivan is included because the work engages substantively with the labour, technology, and political-economic dimensions of specialty-crop automation in the Salinas Valley — dimensions that mainstream agrifood AI reports (USDA AI Strategy, the AI Institutes’ communications, vendor marketing, investment pitches) systematically avoid.

Sullivan’s contribution is based on field visits, interviews, and focus groups in the Salinas Valley from 2021-2025, with primary ethnographic attention to romaine lettuce harvesting and adjacent specialty crops (broccoli, berries). The work is interview-and-fieldwork-based; not peer-reviewed academic publication. Citation count is small; analytical contribution is substantive.

The five substantive findings

1. The “holy grail” framing is industry-side, not neutral. Sullivan documents that the phrase “harvesting of specialty crops remains the elusive ‘holy grail’ of automation” came up in multiple interviews and at conferences, including from automation promoters themselves. The framing is structural: industry-side actors are the ones naming harvest automation as the most difficult problem.

2. The technical limits are concrete, not media-hype. Sullivan’s field research documents specific technical failure modes of automated lettuce harvesters:

• Water-jet-knife approach: “spews mud into the harvest process and ruins otherwise good product.”
• Food-safety concerns: “bacteria buildup at the machine’s joints, where water and material can pool but which are difficult to clean.”
• Robotic finger costs: $500 each, frequently needing replacement due to leaks and cracks in the soft material.
• Mistakes vs. skilled farmworkers: machines make “far more mistakes than a skilled farmworker, such as picking a bad quality lettuce or passing over a ripe one.”

3. The economics do not yet work. “As it stands right now, hiring farmworkers for the romaine lettuce harvest is still cheaper and more reliable than the existing automated harvesters.”

4. A Salinas Valley agriculture automation expert (interviewed by Sullivan April 18, 2025) stated head-lettuce harvest automation is currently at “Zero percent.”

5. The labour dimension is structural, not transitional. “The ongoing attempts to automate romaine lettuce harvesting demonstrate the continued and unquestionable need for careful farmworker labor within so-called digital agriculture. It follows, then, that they need to be paid, treated, and valued according to the value they produce — the Salinas Valley lettuce industry would wilt without farmworkers.”

A farmworker quoted in Sullivan’s focus group (December 2025): “When they tried a machine that cut the lettuce, the lettuce was cut upwards and they tore it apart, the one with hearts [romaine]… . Then I had to fix it. So, people are always going to be needed.”

A roboticist interviewed by Sullivan (February 2025): “You need hands to cut lettuce. Think about a robot with a knife. That’s not happening… . It’s just so delicate. It’s a really hard nut to crack.”

Salinas Valley context (the geographic anchor)

• Monterey County 2024 gross specialty crop production value: just shy of $5 billion.
• Leafy greens production: Salinas Valley accounts for at least 70% of US leafy greens production.
• Romaine lettuce: accounts for almost $1 billion of Monterey County’s $5 billion specialty crop revenue (2024).
• Labour context: California overtime laws (Rural Migration News 30(3), July 2024); H-2A temporary migrant worker program costs; growers responding via automation push AND geographic relocation (Peru, Chile, Mexico).
• Silicon Valley proximity: Salinas Valley is ~1 hour from Silicon Valley. Forbes Salinas Valley AgTech Summit began 2015. Hundreds of subsequent events and partnerships.

The field-pack labour process (the structural finding)

Sullivan documents the field-pack labour process for romaine lettuce — harvested, processed, and packaged in the field, requiring 14 farmworkers across 7 rows per harvest (2 per row). One “cutter” uses a knife to cut the lettuce head and remove outer leaves; the second farmworker of the pair cleans, checks for quality, and seals the packaging directly on a moving platform. The cutting requires within-second skill: “choosing—within seconds—ripe lettuce that is not too young, damaged, or yellowing and then harvesting it without harming the tender, watery vegetable.”

Field-pack is higher-risk for growers (no processing facility for additional inspection/cleaning) but provides unique freshness and quality. This is the labour configuration that automation is attempting to displace — and that Sullivan documents is not yet displaceable at commercial scale.

Adjacent scholarly anchors (Sullivan’s intellectual lineage)

Sullivan cites and extends:

• Madeleine Fairbairn, Zenia Kish, Julie Guthman (UC Santa Cruz) — “Pitching Agri-Food Tech: Performativity and Non-Disruptive Disruption in Silicon Valley,” Journal of Cultural Economy 15(5), 2022, doi:10.1080/17530350.2022.2085142.
• Julie Guthman — The Problem with Solutions: Why Silicon Valley Can’t Hack the Future of Food (University of California Press, 2024).
• Tomás Almaguer — Racial Fault Lines: The Historical Origins of White Supremacy in California (UC Press, 1994).
• Richard Walker — The Conquest of Bread: 150 Years of Agribusiness in California (New Press, 2004).
• Don Mitchell — “Taylorism Comes to the Fields: Labor Control, Labor Supply, Labor Process, and the Twilight of Fordism in California Agribusiness,” Economic Geography 99(4), 2023, doi:10.1080/00130095.2023.2188188.
• William H. Friedland, Amy Barton, Robert Thomas — Manufacturing Green Gold: Capital, Labor, and Technology in the Lettuce Industry (Cambridge UP, 1981). The historical baseline for Salinas Valley labour and technology analysis.
• Madeleine Fairbairn et al. — “Digital Agriculture Will Perpetuate Injustice Unless Led from the Grassroots,” Nature Food 6 (2025), doi:10.1038/s43016-025-01137-8.

This intellectual lineage places Sullivan squarely within the UC Santa Cruz critical food systems scholarship cluster — Guthman, Fairbairn, Walker, Mitchell — and gives her analytical work structural grounding that the agritech industry press generally lacks.

Comparison with Carolan (the other US critical voice)

Michael Carolan (Colorado State University sociology, per units/carolan-colorado-state-critical.md) is the most-substantive US-based sociological critical voice on digital agriculture more broadly. Sullivan is the most-substantive recent critical voice on US specialty-crop AI deployment and farmworker labour specifically. The two complement: Carolan’s work is critical-sociological and engages labour, recognition, and disability dimensions of agricultural AI; Sullivan’s work is ethnographically grounded in the Salinas Valley and engages the labour-substitution question directly through field observation and farmworker voice.

What this unit is doing in the taxonomy

Anchors the specialty-crop AI × farmworker labour × critical-ethnographic voice cell of the matrix. Distinct from:

• Carolan (units/carolan-colorado-state-critical.md) — broader sociological critical voice; tenured academic; Colorado State.
• IPES-Food / FIAN / ETC Group — international critical voices with policy-framework posture.
• Vendor and academic-institute framings of specialty-crop AI — generally optimistic, deployment-focused, technology-positive.

Sullivan’s contribution is the rare substantive empirical counterweight to the “Silicon Valley solves agriculture” narrative, grounded in the specific labour configuration of specialty-crop harvest and the specific failure modes of existing automation.

Why it matters for talks

• The most-substantive recent critical voice on US specialty-crop AI. Joins Carolan as a key US-side critical voice worth referencing in any talk about digital agriculture, labour, or specialty-crop AI.
• The “zero percent” head-lettuce harvest automation finding is a load-bearing concrete datum. Counters vendor and investor pitch decks that conflate research-stage prototypes with commercial deployment.
• The farmworker voice (verbatim quotes from focus groups) is analytically rare in agritech discourse. Sullivan’s fieldwork centers farmworker experience — the structural inversion of vendor framing.
• The intellectual lineage (Fairbairn, Guthman, Walker, Mitchell, Friedland) gives the work academic rigor that the typical agritech trade press lacks.
• The “holy grail” framing as industry-side language is a useful diagnostic: industry actors themselves name the harvest problem as the hardest, while investor pitches often conflate progress on weeding with progress on harvest.

Critical context

• Sullivan is a PhD candidate, not a tenured academic. The work is interview-and-fieldwork-based ethnography published in a media publication (Just Tech / SSRC), not peer-reviewed academic article. Per corpus hygiene: citation volume ≠ substantive contribution. Sullivan’s contribution is substantive; the citation count is small.
• The work focuses on Salinas Valley specifically. Other US specialty-crop labour contexts (Florida berries, Washington apples, Georgia Vidalia onions, California’s Central Valley table grapes) are not surfaced. The farmworker-side critical voice beyond Salinas Valley remains a gap (G-176).
• The interview-based methodology has standard limitations — selection of interview subjects, depth of access, time-frame of fieldwork. Sullivan addresses these in the SSRC essay directly.
• The labour framing is consistent with the broader US specialty-crop labour-replacement narrative — automation as response to H-2A worker costs and California overtime regulations. The structural question is whether automation is targeted at H-2A cost reduction or whether farmworker labour remains structurally necessary. Sullivan’s evidence supports the latter reading.
• The Salinas Valley has substantial existing institutional support for farmworker advocacy (United Farm Workers, various NGOs). Sullivan is not the only voice; the corpus hygiene rule on citation volume vs. contribution still applies.