What's Happening?
Dan Bradbury, an assistant professor of marketing at Western Carolina University’s College of Business, is conducting ongoing research into the benefits of locally operated artificial intelligence (AI). This study examines how running AI models and workflows
on local hardware can provide researchers and organizations with greater control over their data, AI models, research processes, and costs. As AI evolves beyond simple chatbots to autonomous agents capable of complex tasks like web searches, data analysis, and code writing, the question of reliance on external, cloud-based AI providers becomes more pressing. Bradbury's research, supported by a College of Business Summer Research Grant, has demonstrated that Apple silicon technology (M-series chips) can locally run compressed language models with up to 72 billion parameters. This capability offers a viable alternative to sending sensitive data to external model providers, addressing concerns about privacy, institutional policies, and the escalating costs associated with commercial AI services that charge per token.
Why It's Important?
This research is significant because it addresses critical concerns surrounding data privacy, organizational control, and economic efficiency in the rapidly expanding field of artificial intelligence. By demonstrating the feasibility and benefits of local AI, Bradbury's work offers a potential solution for universities, businesses, and researchers who handle sensitive information, such as student submissions, research data, or proprietary company records. Relying on external AI providers can raise privacy obligations, especially concerning regulations like the Family Educational Rights and Privacy Act (FERPA) for educational institutions. Furthermore, the cost model of commercial AI services, which charges based on data input and output, can become prohibitively expensive for intensive AI workflows. Local AI could mitigate these costs, making advanced AI capabilities more accessible and sustainable for organizations. This shift towards local control could empower entities to innovate with AI while safeguarding their data and managing their budgets more effectively, fostering greater digital sovereignty.
What's Next?
Dan Bradbury will continue his research throughout the fall, focusing on refining his findings and developing preliminary results for presentation at the upcoming Appalachian Research in Business Symposium. The project aims to produce both a scholarly article and an open toolkit, which will help researchers and organizations understand the trade-offs between local and externally hosted AI systems and implement transparent workflows for local AI applications in areas like secondary-data research and literature synthesis. This work is part of a broader effort to understand when local AI can serve as a practical alternative or complement to cloud-based solutions. The findings could influence how academic institutions and businesses approach AI adoption, potentially leading to increased investment in local AI infrastructure and the development of more robust internal AI capabilities. Bradbury will also serve on the selection committee for the 2027 College of Business Summer Research Grants, further contributing to the university's research ecosystem.
Beyond the Headlines
The exploration of local AI by Western Carolina University delves into the deeper implications of digital sovereignty, a concept gaining traction as organizations seek to retain control over their technology, data, and infrastructure. This research highlights an ethical dimension: the responsibility of institutions to protect sensitive information when leveraging powerful AI tools. The ability to run AI models locally reduces reliance on third-party vendors, potentially mitigating risks associated with data breaches, vendor lock-in, and opaque algorithmic processes. Culturally, this shift could foster a more decentralized approach to AI development and deployment, empowering smaller organizations and individual researchers who might otherwise be excluded by the high costs and data-sharing requirements of commercial platforms. In the long term, widespread adoption of local AI could reshape the competitive landscape of the AI industry, encouraging the development of more efficient, open-source models and hardware solutions that prioritize user control and privacy.
