Machine Learning and Food Deserts; Risks of AI Development: The Balance Between Regulation and Decentralization

Introduction
My technical project and STS research project are not related. My technical report covers a group project for a machine learning elective, while the STS project dives into ethics in artificial intelligence (AI) development. My group created a machine learning model to identify the causes of food deserts in the state of Virginia. We chose this topic to improve our understanding of basic machine learning techniques while benefiting those in our home state who struggle to find healthy food options. Similarly, I chose to research AI development due to my concern for the direction this technology is headed. Intelligence that can rival or likely surpass human cognitive ability demands a thorough investigation and a delicate creation.

STS Research Project
Focusing on qualitative and ethics-based research, I answered the question – should AI development be monopolized to maintain safety or decentralized to promote competition-based innovation? My sources covered recent events, such as a cumulative UN report on AI standings, as well as events providing historical context. With AI promising to be a power unlike anything else, studying the structure and function of worldwide nuclear arms agreements offers a potential plan for AI oversight. Through this research, I determined that both extremes are dangerous. Unregulated AI leads to powerful agents supporting criminals and terrorists; likewise, a centralized command structure leads to stagnation and inequality. Through diverse, multi-stakeholder oversight committees that support smaller enterprises, prevent monopolization, and uphold complete transparency, my proposal encourages AI innovation while safeguarding the technology from malicious users.

Technical Project
My technical report employed regression and clustering techniques to calculate and visualize the correlation between potential causes and existing food deserts. While this model used simple techniques, they proved effective in identification tasks. Collecting data from Kaggle and other public sources, we performed cleaning and some feature engineering. Due to this project's small computational requirements, Google Colab provided a free platform to create and test our model.

Conclusion
Through these two projects, I gained a deeper understanding of the link between technology and ethics. Tasked with completing a project to help the local community, my technical project group learned of the ease and accessibility of employing machine learning techniques to support real-world problems. Then, for the STS research, I gained a new perspective on the laws, ethics, and best practices for developing new technology. Lessons from these projects will aid my future pursuits as an engineer to balance the efficiency of innovation and the assurance that my work serves all stakeholders.

School of Engineering and Applied Science

Bachelor of Science in Computer Science

Technical Advisor: Briana Morrison, Rich Nguyen

STS Advisor: Richard Jacques

Technical Team Members: Edward Boehling, Matthew Callen