South Carolina's working population includes both some of the most AI-exposed occupational categories in the country and some of the most insulated. The state's workforce infrastructure has more capacity to respond than is generally acknowledged, but the timeline is the issue.
A full treatment of AI's effect on the South Carolina workforce would be a multi-hundred-page study; this chapter is a survey at a level appropriate for a flagship landscape report. SCAIO's published analysis of the gap between what AI can do and what it is doing covers some of this ground in more detail using the Anthropic Labor Market Impacts of AI framework. Readers wanting more granular occupational-exposure analysis are referred there. This chapter sits at a higher altitude.
Anthropic's March 2026 report split occupational AI exposure into two measurements: theoretical exposure (the share of an occupation's tasks AI could plausibly perform) and observed exposure (the share AI is actually being used for in workplaces). For most occupations, theoretical exposure substantially exceeds observed exposure — the gap is closing, but it has not closed.
For South Carolina, the theoretical-exposure picture concentrates in three categories of work:
This category — including general office clerks, customer service representatives, secretaries, executive assistants, bookkeepers, billing clerks, data entry workers, and shipping-and-receiving clerks — is the single largest occupational category in South Carolina by employment count, and it is also the category with the highest observed AI use rate in the Anthropic dataset. Generative AI is already substituting for tasks within these occupations at meaningful scale: drafting correspondence, summarizing documents, scheduling, basic data analysis, customer service triage. The work isn't necessarily disappearing, but it is changing — fewer hours per task, more focus on judgment and exception handling.
These occupations have very high theoretical exposure and very low observed exposure. Clinical and teaching work involves substantial information processing that AI can plausibly assist, but human interaction, professional licensure, and accountability structures slow deployment. MUSC's DAX Copilot pilot (a 20% reduction in clinician documentation time) is an example of AI augmenting rather than replacing — a pattern likely to dominate these sectors for the next several years.
Software developers, data scientists, IT support, and adjacent roles have both high theoretical exposure and rapidly rising observed exposure. Generative AI is changing the productivity of programmers more than the employment of programmers; what counts as "junior developer work" is shifting.
Several large categories of SC employment have meaningfully lower AI exposure on Anthropic's metrics:
The "insulated" framing is partial — every category has internal variation, and the technology is moving fast enough that today's insulation can be tomorrow's exposure. But the pattern at the population level is meaningful: South Carolina's heavy weighting toward manufacturing, construction, and trades partly offsets the high exposure in office and administrative support.
The traditional public-policy response to occupational disruption is workforce training. South Carolina's training infrastructure for this purpose is, by national comparison, in better shape than is sometimes acknowledged.
The state's 16 technical colleges are the most AI-relevant workforce-training infrastructure SC has, full stop. The system serves more than 130,000 students annually across credit and non-credit programs, has long-standing relationships with the major manufacturing OEMs (BMW's apprenticeship and Boeing's training partnerships are the canonical examples), and has the institutional flexibility to ship short stackable credentials in months rather than years. The system has begun adding AI-relevant offerings — IT support with AI-tool exposure, data-analytics certifications, manufacturing-technology programs — but the offerings vary by college and the pace of curriculum update varies.
The technical college system's distinctive workforce-training advantage is that it can compress the time between identifying a workforce gap and shipping a credential program. That advantage is most useful when the gaps are clearly identified. Chapter 8 includes a recommendation that SC consider a more structured AI-credential pathway across the system.
USC, Clemson, MUSC, and the state's other four-year and graduate institutions produce the AI talent that builds and operates the systems described elsewhere in this report. The pipeline is real but small — measured against demand from in-state employers and out-of-state recruiters competing for the same graduates. State-level incentives or programs to retain SC-trained AI talent within the SC economy are an open question.
DEW operates the state's main publicly-funded workforce-development apparatus. AI-specific workforce programming is at an early stage, but DEW has the infrastructure (career centers, employer relationships, federally-funded retraining grants) that makes scaled AI re-skilling possible if and when the workforce demand becomes clear.
ADAPT in SC, profiled in Chapter 2, is the most explicit cross-institutional AI workforce-readiness initiative currently active in the state. Its convening role — bringing universities, technical colleges, employers, and policy actors into the same conversation about AI workforce readiness — is structurally important and underrecognized.
The SC workforce is not a single workforce. Three regional patterns matter:
The region with BMW, the proposed Moc-1 facility, the heaviest tier-1 supplier base, and the deepest existing manufacturing-workforce infrastructure. Re-skilling questions here are mostly about how factory work is changing, not whether it disappears.
MUSC's hiring trajectory, plus defense and cyber work in the Charleston metro, anchors a different workforce composition. The opportunities and pressures here are more in clinical-tech, IT, and data roles than in production roles.
The regions with the highest concentration of the most exposed occupational categories (office and administrative support) and the thinnest existing AI-adjacent training infrastructure. The most consequential retraining work for the state is likely in these regions.
Three things are likely:
The state has more workforce-training capacity than is sometimes acknowledged. It has a manufacturing economy whose AI exposure is heavily about how work is done rather than whether work exists. It has a healthcare and education exposure that is mostly augmentation, not replacement. It has an office and administrative support workforce that is genuinely changing. And it has a technical college system uniquely positioned to address the gaps that emerge.
What it does not yet have is a clearly-named coordinated workforce-AI strategy that connects the technical colleges, four-year institutions, employer base, DEW, ADAPT in SC, and the AI Center of Excellence into a single plan. Chapter 8 returns to this.