A data-driven approach to making better choices | MIT News

Imagine a world where important decisions, such as a judge’s sentencing recommendation, a child’s treatment protocol, or determining who should receive a loan, could be made more reliable with the help of well-designed algorithms. A new MIT economics course is exploring these possibilities.

Class 14.163 (Algorithms and Behavioral Science) is a cross-disciplinary course focusing on behavioral economics, which examines human cognitive capacities and limitations. The course was co-taught by assistant professor of economics Ashesh Rambachan and visiting lecturer Sendhil Mullainathan.

Rambachan’s research focuses on the economic applications of machine learning in decision-making processes in the criminal justice system and consumer lending markets. Mullainathan, who will soon join MIT’s Electrical Engineering and Computer Science and Economics departments as a professor, uses machine learning to understand complex problems in human behavior, social policy, and medicine.

The course aims to understand people scientifically and improve society by enhancing decision-making processes. Rambachan believes that machine-learning algorithms offer new tools for achieving both scientific and applied goals in behavioral economics.

“The course explores how computer science, artificial intelligence (AI), economics, and machine learning can be used to enhance outcomes and reduce bias in decision-making,” Rambachan explains.

Rambachan sees opportunities for AI, machine learning, and large language models (LLMs) to reshape practices in criminal sentencing, healthcare, and other areas where bias may exist.

Students learn to utilize machine learning tools with the objectives of understanding their functions, integrating behavioral economics insights, and identifying areas where these tools can be most effective.

The course also addresses the dangers of subjectivity and bias, emphasizing the importance of understanding biases and mistakes inherent in decision-making processes.

By offering a cross-disciplinary approach to exploring how algorithms can improve problem-solving and decision-making, Rambachan hopes to lay the foundation for redesigning existing systems in various sectors.

Economics doctoral student Jimmy Lin, initially skeptical of the course’s claims, found his perspective transformed as he delved deeper into the material.

Lin praised the course’s emphasis on a “producer mindset” and the importance of asking new questions and creating innovative methods in the intersection of AI and economics.

The course envisions a future where better algorithms can enhance decision-making processes across disciplines, automating the best human choices to improve outcomes while minimizing the worst.

Lin remains enthusiastic about the course’s potential and its impact on the future of research and scientific discovery.