Ten constructive observations about what would help South Carolina's existing AI work connect, scale, and serve its citizens — written from a posture of championing the work already underway, not critiquing it.
This chapter offers ten recommendations. None of them are prescriptions for what the state must do. All of them are observations about what would help — drawn from the assets identified in Chapters 1 through 6, the risks and opportunities laid out in Chapter 7, and the work South Carolina's institutions are already doing.
The recommendations are grouped into five categories: connective tissue (how the existing pieces fit together), capacity-building (how the workforce and citizen literacy keep up), infrastructure (how the physical and procurement substrate of AI gets governed), sector-specific opportunities (where the state could lead), and public trust (how the work stays grounded). Each recommendation is followed by what would help concretely, the relevant SC and peer-state precedents, and a note on who is best positioned to convene the work.
The pieces exist. SCRA's AI Leadership Hub already convenes a public-private-academic cross-section. ADAPT in SC is the state's NSF-funded EPSCoR vehicle. The AI Center of Excellence sits inside the Department of Administration. USC's AI Institute, the Clemson School of Computing, the Clemson–MUSC AI Hub, and the MUSC Center for AI each anchor a piece of the academic capacity. The House Regulations and Administrative Procedures Committee provides a legislative touchpoint. What is missing is a lightweight standing forum that brings their leads into the same room on a regular cadence.
What would help: a quarterly invitation-only convening of approximately 25 leads from the institutions above, on a rotating-chair model with an explicit one-page agenda each meeting, hosted alternately by SCRA, the Department of Administration, and one of the universities. The objective is not new authority. It is shared awareness — what is being deployed where, what is being procured, what is being researched, where collaboration would compound effort.
Peer-state precedent: Virginia's Joint Commission on Technology and Science (JCOTS) provides a long-running model for legislative-technical-academic convening. Pennsylvania's Generative AI Governing Board offers an executive-branch version. Neither maps perfectly to SC's institutional structure, but both demonstrate that lightweight cross-sector forums are durable when their cadence is modest and their remit is clearly bounded.
Best positioned to convene: SCRA, in partnership with the AI Center of Excellence.
What systems are in production, in pilot, or under procurement, by which agencies, with which vendors, for which decisions — at the state and local level. The NIST AI Risk Management Framework provides a usable template for the schema. The OMB M-24-10 federal-agency model demonstrates that public-facing AI use-case inventories are operationally feasible.
What would help: a simple, browsable public inventory listing, for each state agency and participating local jurisdiction, the AI systems in use along with intended purpose, vendor, deployment status, oversight mechanism, and contact for inquiries. Initial scope can be state agencies only; local participation can grow over time. The Department of Administration's existing IT inventory and procurement records provide the substrate; the work is primarily one of consolidation and disclosure standards, not new measurement.
Peer-state precedent: California's Generative AI Use-Case Inventory under EO N-12-23 is the most developed state-level example; New York City's Algorithmic Tools registry is the longest-running municipal model; Pennsylvania's executive-branch generative-AI inventory is the closest peer in scope. The federal M-24-10 inventories provide consistency anchors agencies can borrow from.
Best positioned to convene: Department of Administration / Division of Information Security, with coordination from the AI Center of Excellence.
Three audiences, three different curricula, all aligned to the same vocabulary. Legislators need a curriculum focused on what AI is technically, what policy levers exist, and how to read a vendor's claim. School board members need a curriculum focused on what AI in classrooms looks like, what H.5253 requires, and how to evaluate procurement proposals. Small-business owners need a curriculum focused on practical adoption — what tools cost, what they do well, where they fail, and how to use them responsibly.
What would help: three short, hour-long modules (with self-paced longer versions) developed once and adaptable for each audience, hosted on a publicly accessible platform, and offered in partnership with the institutions that already serve these audiences — the Municipal Association, the SC School Boards Association, the SC Chamber of Commerce, and the Small Business Development Centers network. SCAIO's /learn/ hub is being built to serve as the public-facing component.
Peer-state precedent: Tennessee's AI Advisory Council and Maryland's AI Subcabinet have both produced legislator-facing primers; the National Conference of State Legislatures' AI policy primers are widely used; Utah's Office of AI Policy publishes AI literacy material targeted at small businesses. The novel piece in SC's case is calibrating three audiences to a single vocabulary so that legislators, school board members, and small-business owners can talk to one another about the same concepts.
Best positioned to convene: SCAIO + Center of Excellence + SC ETV, with delivery partners by audience.
South Carolina's Technical College System is the state's most underused AI workforce asset. Sixteen institutions, presence in every county within commuting distance, an existing infrastructure for short, credential-bearing programs, and an established employer-customized training pipeline through ReadySC. Short, stackable certifications for AI-adjacent operational roles could fill a gap that four-year degrees take too long to address.
What would help: a coordinated multi-college curriculum across five or six initial credentials — data labeling and quality assurance, AI-aware customer service operations, MLOps support, applied prompting and workflow integration, model-output review, and AI-aware project management — designed to stack toward a single recognized certificate. ReadySC's existing employer-customized training infrastructure can serve as the channel into the major SC employers (BMW, Boeing, the health systems, the major IT services firms).
Peer-state precedent: Virginia's FastForward credential framework, Texas's High Demand Job Training Program at the Texas Workforce Commission, and Tennessee's TCAT (Tennessee Colleges of Applied Technology) AI-adjacent programs offer well-developed templates. SC's distinctive opportunity is the existing density of the technical college network combined with the employer base.
Best positioned to convene: SC Technical College System, with the SC Department of Employment and Workforce and ReadySC.
Santee Cooper's 50 MW+ large-load rate pilot is a strong starting point, as discussed in Chapter 5. The state's broader opportunity is to extend that transparency principle into a coherent statewide framework that pairs energy-pricing clarity with siting clarity, water-use disclosure, and community-benefit standards — so SC can attract the data-center investment it wants without the externality fights that have surfaced in other states.
What would help: a published framework, jointly developed by Santee Cooper, the Public Service Commission, the Office of Regulatory Staff, and the Department of Commerce, that defines (a) how new large-load AI/data-center customers are evaluated and priced, (b) what water-use, emissions, and community-impact disclosures are required, (c) what community-benefit commitments are expected from developers, and (d) what the public reporting cadence is. The framework does not need to lock in numbers; it needs to lock in process and disclosure.
Peer-state precedent: Virginia's data-center growth (the country's largest concentration in Northern Virginia's "Data Center Alley") has been accompanied by increasingly detailed local zoning frameworks; Iowa's hyperscale incentive structure offers a transparency contrast worth studying; Ohio's recent legislation around data-center electricity costs offers a more recent peer reference. SC's distinctive opportunity is to do this proactively before the buildout reaches Virginia-scale density.
Best positioned to convene: Office of Regulatory Staff, with Santee Cooper, the Public Service Commission, and the Department of Commerce.
State agencies, the technical college system, and the public university system collectively spend significant amounts on enterprise software each year. Coordinated procurement preferences — for vendors with transparent data practices, with SC-based deployment partners, or with workforce commitments — could be a powerful market signal without raising costs.
What would help: a shared state-government AI procurement guidance document, developed by the Materials Management Office in coordination with the AI Center of Excellence, that names (a) the disclosures vendors should make about training data, model evaluation, and deployment limits, (b) the preferences SC agencies will give to vendors that meet specific standards, and (c) the contracting language SC agencies can borrow when negotiating AI contracts. Guidance, not mandates — agencies retain procurement discretion, but the procedural defaults are shared.
Peer-state precedent: California's GenAI procurement guidance and the General Services Administration's federal AI acquisition guidance both provide usable templates. The Government Accountability Office's AI Accountability Framework provides the evaluation backbone. SC's distinctive opportunity is to align procurement preferences with the state's economic-development objectives — preferring vendors with SC-based deployment partners, for instance.
Best positioned to convene: Materials Management Office (Department of Administration), with the AI Center of Excellence.
Linking the universities' AI capacity to the state's manufacturing base — BMW, Boeing, Volvo, Mercedes-Benz Vans, Michelin, Bridgestone, Continental, and the tier-one supplier ecosystem — through a structured industry-academic consortium could cement SC's "industrial AI capital of the Southeast" positioning. Clemson's CU-ICAR and USC's McNair Aerospace Center already do adjacent work; an explicit applied-AI consortium would add a defined remit and a public face.
What would help: a multi-year consortium agreement among Clemson, USC, and an initial group of three to five anchor manufacturers, with shared research priorities, shared workforce pipelines, a small jointly funded applied-research budget, and an annual public showcase. The consortium does not need to start large; a focused initial scope (predictive maintenance, computer vision in quality control, robotics, and supply-chain optimization) is enough to make the alignment legible.
Peer-state precedent: Michigan's UMTRI (University of Michigan Transportation Research Institute) and the Mcity testbed offer a long-running industry-academic-applied template. Indiana's Battery Innovation Center is a more recent example. SC's distinctive opportunity is the scale of the existing automotive and aerospace base combined with two large research universities within an hour of one another.
Best positioned to convene: Clemson and USC jointly, with the SC Department of Commerce.
MUSC's Center for AI and the Clemson–MUSC AI Hub already do much of the underlying work. A voluntary statewide standard for clinical-AI deployment — drawing from MUSC's existing governance, Prisma Health's experience, rural-hospital input, and the Department of Insurance's S.443 oversight role — would let SC lead on a topic where most states are still reactive.
What would help: a published standard, jointly developed by MUSC and the SC Hospital Association, that covers (a) evidence requirements for clinical-AI deployment, (b) bias auditing and demographic-impact review, (c) patient-disclosure practices, (d) governance and oversight structures, and (e) a deployment template that smaller systems can adopt without rebuilding it themselves. The standard would be voluntary — adoption would be a choice, not a mandate — but its existence would create a usable reference.
Peer-state precedent: California's SB 1120 (the "Physician Make Medical Decisions Act," which constrains AI use in coverage decisions) provides a policy bookend; Texas's Health Information Technology Advisory Committee provides a governance template; the Coalition for Health AI (CHAI) provides a national reference framework SC institutions could adapt locally.
Best positioned to convene: MUSC and the SC Hospital Association, with the Department of Insurance and the Department of Health and Environmental Control.
Tracking how South Carolinians actually feel about AI in their schools, clinics, jobs, and government — by region, sector, and demographic — would be one of the highest-leverage ways to keep the policy conversation grounded. The survey instrument is well-developed at the national level; the gap is a sustained, SC-specific instantiation.
What would help: an annual telephone-and-online survey of approximately 1,000 SC residents, with a fielded instrument that covers AI familiarity, AI use, AI concerns, and trust in AI in specific institutional contexts — schools, hospitals, employers, government services, and elections. Results published openly with crosstabs by region and demographic. The Riley Institute at Furman, USC's Institute for Public Service and Policy Research, and the Strom Thurmond Institute at Clemson all have the methodological capacity; one of them, on a rotating basis, could host the work.
Peer-state precedent: The Pew Research Center's national AI surveys provide an instrument template. AP-NORC's recurring AI confidence polling provides a methodology reference. The Strom Thurmond Institute's South Carolina Survey provides a long-running SC-specific public opinion infrastructure SC could build the AI module onto rather than starting fresh.
Best positioned to convene: the Strom Thurmond Institute / Riley Institute / USC IPSPR, on rotation, with SCAIO and the SC Press Association as publication partners.
The work SCAIO does is one piece of this. The Post and Courier, SC Daily Gazette, SC Public Radio, the SC Press Association's network of local papers, and emerging digital outlets are others. The state's information ecosystem will need more, not less, capacity for substantive AI coverage as AI moves from headline territory into the lived experience of South Carolinians.
What would help: dedicated AI-beat reporters or contributors at SC news organizations; small-grant programs for freelance journalists working on SC AI stories; an SC AI journalism fellowship at one of the state's journalism programs; and continued institutional support for nonpartisan research outlets working on the topic. The capacity is built by sustained investment over years, not one-time grants.
Peer-state precedent: ProPublica's local-newsroom partnerships, the Knight Foundation's local-news funding, and Report for America's beat-reporter placements all offer models. The SC Press Association's relationships with the state's news organizations make distribution of any AI-journalism initiative more efficient than in many states.
Best positioned to convene: SC Press Association and the journalism programs at USC, Winthrop, the College of Charleston, Furman, and Coastal Carolina.
South Carolina has more existing capacity than the rest of this report has fully captured. Many of these recommendations describe work that is already underway in pieces; they ask only that the pieces be connected, made visible, and sustained. SCAIO will track which of these recommendations get taken up, by whom, and what they produce — and will revise this chapter in future editions accordingly.