Why AI Academia is Dying (and how to fix it) - Herbie B
A guided tour through the latest and occasionally greatest in the field of artificial intelligence. We'll touch on topics such as large language models, scaling laws, the bitter lesson, overfitting, and Goodhart's law, as well as discuss the industry landscape and the benefits of open science.
About the speaker: Herbie graduated with a CS & Maths degree before coming to Warwick for a masters in Data Analytics, in 2020. He then went to Cambridge for a PhD in AI for Environmental Risk, before switching to the topic of machine learning, working on the intersection of large language models and evolutionary algorithms. He is currently a research lead in the open-source research collective EleutherAI.
Talk by Herbie B.
00:00 - Introduction
00:49 - The Bitter Lesson
11:26 - Research Landscape & Industry Trends
25:22 - What can be done?