MIT’s Schwarzman College of Computing has introduced a groundbreaking interdisciplinary course, "AI and Rationality," designed to challenge the foundational assumptions of modern computing. By merging the technical rigor of computer science with the reflective depth of philosophy, professors Leslie Kaelbling and Brian Hedden are guiding students through the complexities of what it truly means for an artificial system to be "rational."

The curriculum moves beyond standard engineering metrics to investigate whether systems like a Large Language Model can possess true agency or if their decision-making is merely a shadow of human cognitive constraints. Students explore whether irrationality stems from human nature, mathematical frameworks, or the training processes themselves. This shift in perspective encourages future developers to step back from the code and consider the broader context of their work, moving toward a more nuanced understanding of Agentic AI.

In an era of rapid technological shifts, the course focuses on "habits of mind" rather than static doctrines. By formalizing concepts of belief and desire within artificial systems, students learn to navigate a future where the definitions of intelligence and autonomy remain in constant flux. This inquiry is also vital for AI Safety, as aligning machine logic with human values requires a deep philosophical understanding of rational decision-making.

MIT’s Schwarzman College of Computing has introduced a groundbreaking interdisciplinary course, "AI and Rationality," designed to challenge the foundational assumptions of modern computing. By merging the technical rigor of computer science with the reflective depth of philosophy, professors Leslie Kaelbling and Brian Hedden are guiding students through the complexities of what it truly means for an artificial system to be "rational."

The curriculum moves beyond standard engineering metrics to investigate whether systems like a Large Language Model can possess true agency or if their decision-making is merely a shadow of human cognitive constraints. Students explore whether irrationality stems from human nature, mathematical frameworks, or the training processes themselves. This shift in perspective encourages future developers to step back from the code and consider the broader context of their work, moving toward a more nuanced understanding of Agentic AI.

In an era of rapid technological shifts, the course focuses on "habits of mind" rather than static doctrines. By formalizing concepts of belief and desire within artificial systems, students learn to navigate a future where the definitions of intelligence and autonomy remain in constant flux. This inquiry is also vital for AI Safety, as aligning machine logic with human values requires a deep philosophical understanding of rational decision-making.