Focus on Fine-Tuning: Understanding the Pathways for Shaping Generative AI – Prof. Paul Ohm (Georgetown)

In this episode of the CLE Vlog & Podcast series, Prof. Paul Ohm (Georgetown University) and Prof. Filippo Lancieri (Georgetown University, formerly ETH Zurich) discuss Prof. Ohm's study on the pathways for shaping generative AI, with a special focus on fine-tuning. In his study, Prof. Ohm explains the differences between pretraining, fine-tuning, in-context learning, and input and output filtering, highlighting why in particular fine-tuning presents the best mix of cost versus capability for shaping the outputs of generative AI. He also analyzes why fine-tuning is especially well-suited for legal interventions, through a case study on fixing biased models. Finally, Paul Ohm evaluates the market and competition implications of focusing on fine-tuning, such as considerations about the small number of market actors who can afford the costs of the pretraining stage. In this vlog episode, Paul Ohm discusses his study in detail with Filippo Lancieri. Paper References: Paul Ohm – Georgetown University Focusing on Fine-Tuning: Understanding the Four Pathways for Shaping Generative AI Columbia Science and Technology Law Review, 25, 214–240 (2024) https://journals.library.columbia.edu/index.php/stlr/article/view/12762/6286 Audio Credits for Trailer: AllttA by AllttA https://youtu.be/ZawLOcbQZ2w