Adam Gleave

Adam Gleave

Founder & CEO at FAR AI

FAR AI

Biography

I am the CEO and co-founder of FAR AI, an alignment research non-profit working to incubate and accelerate new alignment research agendas. I received my PhD from UC Berkeley under the supervision of Stuart Russell. I was fortunate to be part of the Center for Human-Compatible AI during my PhD and funded by an Open Philanthropy Fellowship. Previously, I’ve spent time at DeepMind working with Jan Leike and Geoffrey Irving, and Cambridge working with Zoubin Ghahramani and Christian Steinruecken. Please see my CV for a more comprehensive list of my prior experience.

Interests
  • Artificial Intelligence
  • Deep RL
  • Beneficial AI
Education

  • PhD in Artificial Intelligence, 2022

    UC Berkeley

  • MPhil in Advanced Computer Science, 2016

    University of Cambridge

  • BA (Hons) in Computer Science, 2015

    University of Cambridge

Publications

Tony Wang, Adam Gleave, Nora Belrose, Tom Tseng, Joseph Miller, Michael Dennis, Yawen Duan, Viktor Pogrebniak, Sergey Levine, Stuart Russell (2022). Adversarial Policies Beat Superhuman Go AIs. ICML.

PDF Cite Code Project

Adam Gleave, Mohammad Taufeeque, Juan Rocamonde, Erik Jenner, Steven H. Wang, Sam Toyer, Maximilian Ernestus, Nora Belrose, Scott Emmons, Stuart Russell (2022). imitation: Clean Imitation Learning Implementations. arXiv.

PDF Cite Code

Erik Jenner, Herke Van Hoof, Adam Gleave (2022). Calculus on MDPs: Potential Shaping as a Gradient. arXiv.

PDF Cite

Pavel Czempin, Adam Gleave (2022). Reducing Exploitability with Population Based Training. arXiv.

PDF Cite

Adam Gleave, Sam Toyer (2022). A Primer on Maximum Causal Entropy Inverse Reinforcement Learning. arXiv.

PDF Cite

Adam Gleave, Geoffrey Irving (2022). Uncertainty Estimation for Language Reward Models. arXiv.

PDF Cite

Joar Skalse, Matthew Farrugia-Roberts, Stuart Russell, Alessandro Abate, Adam Gleave (2022). Invariance in Policy Optimisation and Partial Identifiability in Reward Learning. ICML.

PDF Cite

Erik Jenner, Adam Gleave (2021). Preprocessing Reward Functions for Interpretability. Cooperative AI Workshop at NeurIPS.

PDF Cite Code

Antonin Raffin, Ashley Hill, Adam Gleave, Anssi Kanervisto, Maximilian Ernestus, Noah Dormann (2021). Stable-Baselines3: Reliable Reinforcement Learning Implementations. JMLR.

PDF Cite Code Project Blog

Adam Gleave, Michael Dennis, Shane Legg, Stuart Russell, Jan Leike (2021). Quantifying Differences in Reward Functions. ICLR (Spotlight Paper).

PDF Cite Code Slides Video OpenReview Blog

Pedro Freire, Adam Gleave, Sam Toyer, Stuart Russell (2020). DERAIL: Diagnostic Environments for Reward And Imitation Learning. DeepRL Workshop at NeurIPS.

PDF Cite Code

Eric J. Michaud, Adam Gleave, Stuart Russell (2020). Understanding Learned Reward Functions. DeepRL Workshop at NeurIPS.

PDF Cite Code

Adam Gleave, Michael Dennis, Cody Wild, Neel Kant, Sergey Levine, Stuart Russell (2020). Adversarial Policies: Attacking Deep Reinforcement Learning. ICLR.

PDF Cite Code Project Slides Video OpenReview Poster Blog

Aaron Tucker, Adam Gleave, Stuart Russell (2018). Inverse Reinforcement Learning for Video Games. DeepRL Workshop at NeurIPS.

PDF Cite Code

Adam Gleave, Oliver Habryka (2018). Multi-task Maximum Causal Entropy Inverse Reinforcement Learning. GoalsRL Workshop at ICML.

PDF Cite Code Slides

Sören Mindermann, Rohin Shah, Adam Gleave, Dylan Hadfield-Menell (2018). Active Inverse Reward Design. GoalsRL Workshop at ICML.

PDF Cite

Adam Gleave, Christian Steinruecken (2017). Making Compression Algorithms for Unicode Text. Data Compression Conference.

PDF Cite Code Extended Abstract Master's thesis

Ionel Gog, Malte Schwarzkopf, Adam Gleave, Robert Watson, Steven Hand (2016). Firmament: Fast, Centralized Cluster Scheduling at Scale. OSDI.

PDF Cite Code Project

Opinions

Some of my opinions are best expressed in formats other than an academic paper. Here are a few of my more notable interviews and essays:

More people getting into AI safety should do a PhD (March 2024).

Writing Beautifully in LaTeX (August 2020). Design patterns for composing LaTeX documents.

Careers in Beneficial AI Research (July 2020). A guide for those interested in AI research careers that have a social impact, with a focus on graduate school.

Conversation with AI Impacts (August 2019). My reasons for being (cautiously) optimistic about the future of AI.