Last Week in AI The De-democratization of AI: Deep Learning and the Compute Divide in Artificial Intelligence Research with Nur Ahmed
Dec 5, 2020
Nur Ahmed, a Strategy PhD candidate at Ivey Business School and Research Fellow at ScotiaBank Digital Banking Lab, dives into the de-democratization of AI. He discusses the alarming concentration of power in AI research and the impact of corporate interests on innovation. Highlighting the widening compute divide, Nur calls attention to the diminishing contributions from non-elite institutions and advocates for more equitable access to AI technology. He emphasizes the need for policy interventions to ensure diverse voices in the AI landscape.
AI Snips
Chapters
Transcript
Episode notes
AI Democratization Concern
- Experts worried about AI power concentrating in a few companies' hands.
- No systematic evidence supported this concern, prompting this study.
Research Methodology
- Researchers analyzed major computer science conference papers to quantify corporate influence.
- They compared AI and non-AI conferences using CSranking.org data and a synthetic control method.
Industry Co-authorship Surge
- Industry co-authorship in AI conference papers drastically increased, unlike non-AI conferences.
- For example, NeurIPS saw a jump from 20% to 40% in the last decade.