Sign In

Communications of the ACM

ACM Opinion

Connecting Large Language Models to Human Values

View as: Print Mobile App Share:
AI researcher Connor Leahy

EleutherAI co-founder Connor Leahy is an AI researcher working on understanding large machine-learning models and aligning them to human values.

EleutherAI is a decentralized grassroots collective of volunteer researchers, engineers, and developers focused on artificial intelligence (AI) alignment, scaling, and open source AI research. In an interview, co-founder Leahy discusses the organization's flagship project: the GPT-Neo family of models designed to match those developed by OpenAI as GPT-3.

Topics include replicating GPT-2/GPT-3, deciding whether to relese GPT-2 weights, life after GPT-2, GPT-3 and the start of EleutherAI, superhuman AI, AI alignment and releasing powerful models, AI risk and research norms, and more.

From The Gradient
Listen to Podcast


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account