Artificial General Intelligence

4.0 credits

Many companies are working towards creating artificial general intelligence (AGI), i.e., systems capable of accomplishing any human intellectual task, and multiple companies aspire to create such systems within the next decade. WHile large language models (LLMs) do not yet have these capabilities, they exhibit "sparks" of AGI. In this course we review the concept of AGI and how we might assess a system to understand if it is an AGI. We will study the gaps in state-of-the-art systems and what would be needed for a system to be considered an AGI. We will also study AI ethics and safety, the socioeconomic implications of AGI, the likelihood of an intelligence explosion (i.e., the singularity), and existential risk due to AGI. Coursework includes readings, student presentations, and a project. Students are expected to be familiar with Python, one or more deep learning toolboxes, deep learning, and machine learning more broadly. Students should have a solid understanding of large language models, e.g., GPT-based causal transformers and similar approaches, backpropagation, multi-layer perceptrons, transformers, convolutional neural networks, and neural network fine-tuning.