CS50x 2026 - Artificial Intelligence

This lecture on artificial intelligence (AI) explores the underlying principles and recent advancements in generative AI. It begins with a demonstration of AI's capabilities through image and text recognition, highlighting the increasing difficulty in distinguishing AI-generated content from reality. The lecture then delves into the architecture behind CS50's virtual rubber duck, emphasizing prompt engineering and the use of system and user prompts. It further covers machine learning concepts, including reinforcement learning, supervised learning, and neural networks, using examples like teaching a robot to flip pancakes and AI playing Breakout to illustrate these principles. The lecture concludes by discussing large language models (LLMs) and the challenges of AI hallucinations, using a poem to highlight AI's imperfections.