The Blind Spot in the AI Jobs Panic: Are We Forgetting Blue-Collar Workers?

As I scroll through my Instagram feed or open my inbox on any given day, I always encounter some version of the same list: the jobs that AI is coming for. Data analysts, paralegals, junior software engineers, financial analysts, content writers. The rankings vary, but the pattern is amazingly consistent — the jobs described as most vulnerable are almost exclusively white-collar, knowledge roles: positions that tend to require higher education, technical expertise or intangible knowledge work.

This is not a coincidence, as it reflects a real, well-documented shift in where AI is having its most immediate impact. But it also reflects a blind spot: one that could have serious consequences if left unexamined. While the media, academia, and policymakers fixate on the displacement of lawyers and accountants, a parallel transformation is underway on factory floors, logistics warehouses, and across the physical economy. I keep having this question in my head: are blue-collar workers being left out of the conversation? When displacement does reach them (as it increasingly will) they will have far fewer cushions to fall back on. So why aren’t we talking about that?

Physical AI is Already Here

The popular narratives frame AI job displacement as a software problem: LLMs replacing tasks done at desks, laptops, over emails. But this framing ignores a rapidly advancing frontiers of physical AI — that is, intelligent systems embedded in robots that can perceive and navigate the real world.

The investment activity alone signals how seriously the industry is taking this. Field AI, a robotics AI company that provides a software platform for mobile robots without maps, GPS, or human supervision, raised nearly $400 million to build software that enables robots to operate autonomously in energy, logistics, and construction spaces [1]. Skild AI, a Pittsburgh-based robots software maker, raised around $1.4 billion at a $14 billion valuation earlier this year with the motto “Any robot. Any task. One brain,” which signals a bet that AI-powered robots can generalize across physical tasks the way LLMs generalize across text [2].

Big tech is not watching from the sidelines. Google DeepMind has invested heavily in embodied AI, with their Robotics Transformer 2 (RT-2), a vision-language model that interprets natural language commands and visual inputs to perform real-word physical tasks [3]. Amazon rolled out Sparrow in 2022, an intelligent robotic system that uses vision and AI to detect, select, and handle products in its fulfillment inventory. The company also deploys Digit, which is a bipedal robot developed by Agility Robotics, to pick up containers and transport them to conveyor belts across its warehouses. These are tasks that were, until recently, exclusively human.

These are not just some speculative demos. They are deployed, revenue-generating systems operating inside the supply chains and warehouses that employ millions of blue-collar workers. The question is not whether physical AI will affect these workers. Rather, it is whether anyone in power will be paying attention when it does.

The Class Blind Spot

I think there is an obvious reason why blue-collar displacement has received less attention: physical AI has, until recently, lagged far behind the capabilities of software-based AI. Blue-collar jobs rely on physical ability, dexterity, and spatial reasoning — qualities that robots have historically struggled to replicate at the human level. It makes sense, then, that early AI displacement discourse focused on cognitive tasks. LLMs arrived first, and they arrived capable of analyzing, drafting, coding, and summarizing at a level that threatened knowledge work.

But the assumption that physical jobs are safely beyond automation’s reach is becoming harder to sustain. According to Michigan Journal of Economics, the global market for humanoid robots is expected to grow exponentially over the next decade, potentially reaching $38 billion by 2035 [4]. The robots being deployed today are not the rigid, single-task machines of previous decades. They are AI-guided systems that can generalize across physical environments with agency.

More importantly, even if white-collar displacement is the more immediate concern, the consequences of blue-collar displacement are likely to be more severe. Only about 42% of blue-collar workers hold a college degree, making it significantly harder for them to pivot into alternative employment when their roles disappear [4]. A displaced junior lawyer can retain, leverage her professional network, and use her credentials to move laterally into adjacent roles. An assembly line worker in a manufacturing company does not have the same options. Blue-collar workers simply have fewer cushions in general: thinner networks, weaker social safety nets, relative lack of opportunity to upskill.

This asymmetry has significant implications. The Brookings Institution has noted that while most AI exposure studies focus on white-collar occupations, they “fail to capture core non-technological factors that influence which workers would experience the most severe welfare costs” from displacement [5]. This means that the workers receiving the most attention are not necessarily the ones who will suffer the most. Rather, the workers receiving the least attention might be.

Is Policy Forgetting an Entire Class of Workers?

The policy response to AI-driven job displacement has largely mirrored the media narrative: it centers on knowledge workers. Retraining programs, workforce development initiatives, and labor market monitoring efforts seem to be heavily designed around digital literacy, AI fluency, and transitions into tech-adjacent roles. There seems to be a lack of policy infrastructure being built for workers who will lose jobs not to a chatbot, but to a robot on a factory floor. This is a critical gap. Retraining a displaced paralegal and retraining a displaced warehouse picker require fundamentally different interventions: different skill developments, different institutional partnerships, different timelines, and different levels of support. Focusing on designing policy around one while paying less attention to the other means that when physical AI displacement accelerates, the workers in its path will be left without a floor to land on.

Implications: What Happens Next

The consequences of this blind spot, if left unaddressed, are significant.

For workers, those who will most likely to be hurt by physical AI are the same ones who already lost out when factories closed and jobs moved overseas. Unlike white-collar workers who have more opportunities to be upskilled, reskilled, and absorbed into adjacent roles, many displaced manufacturing and logistics workers face permanent labor force exit. The negative spiral (job loss, income decline, community disinvestment) is well-known from previous waves of automation. Physical AI threatens to accelerate it.

For policy, the gap between where disruption is headed and where policy attention is focused needs to close. This means designing AI workforce programs that address physical job transitions specifically. A rollout of targeted support for workers in industries where physical AI and robot adoption is accelerating and apprenticeships in robot supervision and maintenance seems necessary and quite urgent than what many think.

Conclusion

The AI jobs conversation is not wrong. White-collar displacement is real and deserves the attention it currently receives. However, I believe it is incomplete. By centering on knowledge workers, we are designing policy, media narratives, and public awareness for one kind of displaced worker while leaving another kind unprepared.

The physical AI wave is not distant. The robots are already in warehouses. The AI brains are already being funded at billion-dollar valuations. The question is whether we will build the infrastructure, policy, and support network to protect the workers before displacement arrives, or only after it is too late.

Footnote

1. Primack, Dan. “FieldAI Raises Over $400 Million to Make Robot ‘Brains.’” Axios, 20 Aug. 2025, www.axios.com/2025/08/20/fieldai-raises-over-400-million-to-make-robot-brains.
2. Szkutak, Rebecca. “Robotics Software Maker Skild AI Hits $14B Valuation.” TechCrunch, 14 Jan. 2026, techcrunch.com/2026/01/14/robotic-software-maker-skild-ai-hits-14b-valuation/.
3. Chebotar, Yevgen, and Tianhe Yu. “RT-2: New Model Translates Vision and Language into Action.” Google DeepMind, 28 Jul. 2023, deepmind.google/blog/rt-2-new-model-translates-vision-and-language-into-action/.
4. Gentry, Isabella. “AI on the Job Industry: How Blue-Collar and White-Collar Workers Are Impacted.” Michigan Journal of Economics, 13 Mar. 2026, sites.lsa.umich.edu/mje/2026/03/13/ai-on-the-job-industry-how-blue-collar-and-white-collar-workers-are-impacted/.
5. Manning, Sam, et al. “Measuring US Workers’ Capacity to Adapt to AI-Driven Job Displacement.” Brookings Institution, 21 Jan. 2026, www.brookings.edu/articles/measuring-us-workers-capacity-to-adapt-to-ai-driven-job-displacement/.