Siobahn Day Grady Confronts AI Education's Funding Divide

Siobahn Day Grady has built the first AI research institute at an HBCU, despite a funding system that gives these institutions less than one percent. An IEEE Spectrum profile presents her work as an ambitious effort to make artificial intelligence accessible across North Carolina Central University. The sharper story is whether that model can survive the financial imbalance it seeks to overcome.

Grady launched NCCU’s Institute for Artificial Intelligence and Emerging Research, or IAIER, in January 2025. It has since engaged more than 2,800 students, faculty members, and community residents, according to the AI literacy profile. Its programs reach beyond computer science into social work, health care, digital archives, and other fields.

The institute’s early reach makes a case for treating AI literacy as general education. Yet its dependence on grants and technology companies exposes a harder constraint. Universities with fewer research dollars must prepare students for an AI-shaped labor market while competing against institutions with deeper faculty, computing, and fundraising resources.

That is the central conflict behind Grady’s work. A university can make AI education universal inside the classroom, but access to the infrastructure supporting that education remains highly unequal.

Grady Turned an AI Lab Into a Campus-Wide Institute

IAIER changed NCCU’s AI strategy from a specialist research project into a university-wide workforce and education program.

Grady’s plan did not begin with the institute. She established the Laboratory for Artificial Intelligence and Emerging Research at NCCU in 2020. The laboratory gave students opportunities to work on projects involving machine learning, human-computer interaction, health care, and autonomous systems.

That structure still centered on a research group. It could serve participating students and support selected projects, but it could not reach every department. Grady began considering a broader model when a Google grant opportunity emerged in 2024.

The resulting institute received $1 million from Google.org. NCCU describes IAIER as a center for interdisciplinary education, research, and responsible AI development. Its institute overview also identifies it as the first HBCU-based AI institute in the United States.

The university’s trustees formally approved IAIER on December 17, 2025, after an institutional review. That approval gave the organization a standing beyond the original faculty laboratory. It also positioned the institute to coordinate academic programs, partnerships, research grants, and community training.

This distinction matters because AI literacy is not the same as training computer scientists. AI literacy means understanding what AI systems can do, where they fail, and how their outputs affect decisions. It also includes the ability to question data sources, identify risks, and use appropriate tools within a specific discipline.

A nursing student and an information-science student will not use AI in identical ways. Neither will a social worker, archivist, teacher, or business analyst. A campus-wide institute can connect shared foundations with the professional standards of each field.

IAIER’s seed-grant program illustrates that approach. The institute awarded grants of up to $10,000 to 11 projects in its first reported cohort. Those projects covered social work, health care, digital archiving, information science, and other areas.

One project is developing an AI-supported simulation lab for social work students. Students can practice client interactions without treating the system as a replacement for supervised field experience. The scenario turns AI education into a question about professional judgment, not merely prompt construction.

That is a more demanding goal than showing students how to operate a chatbot. It requires faculty members to decide when AI improves learning, when it introduces unacceptable uncertainty, and how students should document its use.

IAIER also organizes events involving companies such as Deloitte, FICO, Anthropic, IBM, and Google. These connections expose students to current workplace expectations and give faculty a view into changing practices outside academia.

Industry involvement brings its own tension. Companies develop AI products faster than universities can revise curricula, approve programs, and train instructors. Universities therefore need outside knowledge, but they cannot let vendors define AI literacy entirely around their own platforms.

Grady’s institute sits directly inside that tension. It uses corporate support to broaden participation while framing AI education as an interdisciplinary academic responsibility. The institute’s next test is whether that balance can hold as programs and demand expand.

IEEE Spectrum Shows Why AI Literacy Is Becoming General Education

The IEEE Spectrum account makes a clear case that AI literacy now belongs across the curriculum, not inside one technical department.

NCCU does not yet have a dedicated computer science program, although the university is developing a computer science major and an AI minor. That absence might appear to be a disadvantage. It has also pushed IAIER toward a model that does not use computer science enrollment as the gatekeeper.

Every NCCU freshman now encounters introductory AI training through UNIV 1100, a required university education course. The program began with a first-year class of nearly 1,700 students during fall 2025, according to NCCU’s required AI module.

The module uses IBM SkillsBuild, an online learning platform that offers foundational technology training and credentials. Students complete the material regardless of their intended major. NCCU says they can also gain access to mentoring and internship pipelines involving IBM and other partners.

Embedding the module inside a required course changes who receives the training. An optional workshop usually attracts students who already feel comfortable with technology. A universal requirement reaches students who might otherwise conclude that AI is irrelevant to their work.

That design reflects Grady’s argument that people can no longer separate digital systems from ordinary professional life. Hiring managers increasingly expect graduates to evaluate AI-assisted work, communicate with automated systems, and recognize unreliable outputs. Those expectations now extend well beyond software development.

However, foundational prompting cannot be the endpoint. A student who knows how to request a summary does not automatically understand hallucinations, privacy, copyright, or embedded bias. Hallucination is the production of plausible but unsupported information by a generative model.

AI literacy must therefore include verification. Students need to compare outputs with original materials, disclose meaningful AI assistance, and understand when confidential information should stay outside third-party systems.

This work also has a knowledge-management dimension. Students and researchers need reliable ways to organize source material before asking AI to interpret it. A personal AI knowledge base can support that process when it preserves citations and keeps claims connected to evidence.

Faculty development becomes equally important. An institution cannot establish common AI expectations if instructors receive no time or guidance to redesign assessments. IAIER has worked with faculty development teams to help educators integrate AI into their courses.

Integration does not mean using the same tool in every classroom. A history instructor might focus on source verification. A social work instructor might examine consent and professional boundaries. A health researcher might concentrate on privacy, validation, and the consequences of false predictions.

This disciplinary approach gives IAIER a stronger rationale than a generic technology center. It treats AI as a layer crossing many forms of work while recognizing that each profession has different risks.

The model also creates pressure for other universities. Institutions that restrict AI instruction to technical electives risk graduating students with uneven preparation. Wealthier universities can respond by hiring faculty and creating specialized centers, but smaller institutions face harder choices.

NCCU has chosen breadth first. It is giving every freshman a baseline while building advanced research and degree options behind that foundation. The approach is practical, but its quality will depend on continued faculty support and regular curriculum updates.

Universal AI Access Meets an Unequal Research System

Grady’s model challenges a structural mismatch: HBCUs are expected to prepare AI talent without receiving an equal share of research investment.

Historically Black colleges and universities represent 3.2 percent of four-year degree-granting institutions. Yet they received only 0.91 percent of federal higher-education research and development expenditures in fiscal 2023.

The gap is not confined to one year. Since 2018, HBCUs have received an average of 0.87 percent of federal academic research funding, according to published federal R&D data.

The distribution across agencies adds another layer. Seventeen of the 43 federal agencies providing university research money in 2023 awarded nothing to HBCUs. That means 40 percent of participating agencies made no HBCU research allocation.

Two major funding sources also distributed disproportionately small shares. The Department of Health and Human Services directed 0.54 percent of its academic research funding to HBCUs. The Department of Defense directed 0.40 percent.

Those figures matter for AI because research capacity compounds. A federal award can fund graduate researchers, computing access, administrative staff, and pilot studies. Those assets help an institution compete for the next grant and attract additional partners.

Limited funding produces the reverse effect. Faculty members spend more time assembling proposals with less administrative support. Institutions struggle to sustain laboratories between grants. Students receive fewer paid research opportunities that can lead to graduate study or technical employment.

The problem is not a lack of relevant talent. HBCUs have long served students who remain underrepresented in science and engineering professions. Their involvement is especially important when AI systems affect communities historically excluded from technology design.

A researcher’s background does not guarantee fair technology. However, excluding institutions and communities from research reduces the range of questions being asked. It also narrows the pool of people positioned to challenge questionable assumptions.

IAIER’s interdisciplinary grants show how broader participation can change the research agenda. A social work faculty member approaches AI differently from a model developer. An archivist sees preservation, provenance, and representation problems that a commercial product team might overlook.

The institute also gives students a place to investigate those questions before entering the workforce. Research experience can teach them to frame problems, handle evidence, test claims, and communicate uncertainty. Those abilities remain valuable even as specific AI tools change.

Corporate grants can help close immediate gaps, but they do not replace sustained public investment. A company can change its philanthropic priorities, reorganize a program, or end a partnership. Universities must support students and faculty across longer cycles.

Government funding can also shift. Federal research budgets face political negotiation, while diversity-focused programs face additional scrutiny. A one-time allocation can create activity without guaranteeing staffing or operations after the award ends.

The federal government announced a $500 million one-time investment in HBCUs and institutions chartered by Native American tribal governments in September 2025. That commitment offered near-term support, but the one-time structure did not resolve the recurring research gap.

For IAIER, the funding divide turns growth into a risk. Success attracts more students, faculty projects, community requests, and potential partners. Each new program adds coordination and support needs that a temporary grant might not cover.

Grady has identified funding as the institute’s largest sustainability barrier. That assessment complicates the upbeat story created by high participation. Reaching thousands of people proves demand, but it does not prove that the operating model can finance itself.

The primary opponent is therefore not another university or AI laboratory. It is the gap between universal access as a promise and unequal research capacity as the reality.

Industry Partnerships Expand Access but Cannot Set the Curriculum

Technology partnerships give NCCU valuable resources, yet the university must retain control over what responsible AI education means.

IAIER has worked with Google and IBM on training and credentials. It has also hosted sessions involving employers and AI companies. These relationships connect students with tools, researchers, professional networks, and potential career pathways.

The institute hosted the first OpenAI Academy Summit held at an HBCU. The event brought 444 participants from more than 40 institutions to NCCU, according to the IEEE Spectrum account. The official HBCU summit included discussions about education, workforce preparation, entrepreneurship, and responsible adoption.

That network value is significant. Students at well-funded universities often gain informal access to visiting researchers, alumni, and recruiters. Events hosted at NCCU can reduce the distance between HBCU students and the organizations shaping AI employment.

Credentials can also make newly acquired skills visible. A student without a technical major may benefit from a recognized certificate that signals foundational training. However, credentials are only useful when employers understand what they measure.

There is a risk that AI literacy becomes a collection of vendor badges. A badge can verify course completion without showing whether a student can challenge an output, protect sensitive data, or identify an unsuitable application.

Vendor-centered education can also age quickly. Product interfaces, model names, and subscription terms change. Students need transferable concepts that remain useful when the preferred tool changes.

Universities must therefore distinguish product familiarity from durable literacy. Product familiarity helps students become productive with current systems. Durable literacy helps them evaluate future systems and recognize recurring problems.

The same principle applies to faculty research. Corporate support can fund experiments and offer technical access, but research questions should not depend entirely on a company’s commercial roadmap. Independent inquiry is essential when scholars study bias, labor effects, privacy, or system failures.

IAIER’s interdisciplinary structure provides some protection against narrow vendor thinking. Its projects begin with problems inside fields such as social work and archiving. The technology serves the research question, rather than defining it.

Still, the institute’s reliance on external partnerships creates a legitimate uncertainty. Publicly reported participation numbers do not reveal how many students complete sustained training. They also do not show whether graduates apply these skills effectively after leaving NCCU.

There is little published longitudinal evidence about employment outcomes, research productivity, or changes in faculty practice. That absence is understandable for a young institute, but it limits broad claims about the model’s effectiveness.

The relevant question is not whether IAIER has already solved unequal AI access. It has not had enough time, funding, or published outcome data to support that conclusion. The question is whether its early design creates measurable progress that other institutions can reproduce.

A useful evaluation would separate event attendance from deeper engagement. It would track course completion, research participation, internships, credentials, graduate placements, and student confidence in verifying AI outputs.

It should also examine who benefits inside the university. Universal freshman training creates a broad starting point, but advanced opportunities can still concentrate among students with more time, prior experience, or faculty connections.

Faculty capacity presents another limit. Seed grants can launch promising projects, but researchers need time, technical support, and follow-on funding. Otherwise, pilot programs disappear after producing an initial demonstration.

None of these concerns invalidate Grady’s strategy. They define what evidence is still needed. The institute has shown that an HBCU can organize a visible, cross-campus AI initiative. It has not yet shown whether the model remains stable after early grants end.

Three Signals Will Show Whether the Model Can Last

IAIER’s future will depend on sustained infrastructure, measurable student outcomes, and replication beyond one campus.

The first signal is NCCU’s dedicated physical space for the institute. Grady told IEEE Spectrum that IAIER planned to open its first dedicated campus facility in fall 2026. NCCU has described a building of roughly 5,000 gross square feet for learning and research.

A physical center would give students and faculty a consistent place for projects, mentoring, training, and collaboration. It would also demonstrate that the university is embedding IAIER into its operations instead of treating it as a temporary grant program.

The facility alone will not establish sustainability. The important details will include staffing, equipment, computing access, operating funds, and the number of programs it can support. A building without recurring resources would weaken the institute’s broader claim.

The second signal is the launch and adoption of NCCU’s planned computer science major and AI minor. These programs would connect universal introductory training with deeper academic pathways.

Enrollment and completion data will matter more than the announcement. Strong demand would show that broad AI exposure encourages students to pursue advanced study. Weak progression might indicate that introductory training needs stronger advising, prerequisites, or financial support.

Outcomes should extend beyond technical majors. IAIER’s distinctive proposition is that AI belongs across disciplines. Researchers should therefore watch how many departments create AI-related coursework, how faculty use seed grants, and whether interdisciplinary projects continue beyond pilot stages.

The third signal is whether NCCU’s approach becomes reusable elsewhere. Grady has said she wants to create a framework that can help HBCUs and other universities establish similar programs.

Replication would strengthen the case that IAIER represents more than a single successful leader or grant. It would also test which elements transfer across institutions with different curricula, student populations, and industry relationships.

A repeatable framework would need clear minimum requirements. Universities would need guidance on faculty development, responsible-use policies, student assessment, research governance, employer involvement, and sustainable funding.

It would also need room for local priorities. A rural institution may focus on agriculture or health access. An urban university may emphasize public administration, transportation, or community entrepreneurship.

Over the next several months, these three signals should clarify the institute’s trajectory. A staffed facility would strengthen its institutional base. Growing academic pathways would demonstrate deeper student demand. Adoption by other universities would support Grady’s claim that the model can travel.

Failure on any one signal would not erase the institute’s early work. However, delayed facilities, weak progression, or absent replication would show where grant-backed enthusiasm meets organizational limits.

The IEEE Spectrum profile captures a leader pushing higher education toward a broader definition of AI readiness. Grady’s most consequential idea is not that every student needs to become a programmer. It is that every student needs enough understanding to question, direct, and verify automated systems.

That idea carries practical consequences for students, employers, and universities. Employers should look beyond tool-specific badges and ask candidates how they evaluate AI-assisted work. Universities should measure sustained learning, not just attendance. Students should treat verification and domain judgment as core skills.

The story also places responsibility on funders. It is difficult to demand a more representative AI workforce while maintaining a research system that directs less than one percent of federal academic R&D funding to HBCUs.

Readers should watch what happens after the initial attention fades. Does IAIER gain recurring support, publish meaningful outcomes, and help other campuses build comparable programs? Those results will determine whether universal AI literacy becomes an institutional commitment or remains dependent on exceptional leaders chasing temporary funds.

Grady has already established the demand. The next step belongs to universities, employers, public agencies, and technology companies. They must decide whether AI literacy is a shared educational priority and fund it accordingly.