Skip to content

cv.md

Latest commit

 

History

History
64 lines (41 loc) · 5.68 KB

cv.md

File metadata and controls

64 lines (41 loc) · 5.68 KB
Raw

Joe Edelman

Berlin · nxhx.org · joe@meaningalignment.org @edelwax · GitHub · Google Scholar

Current

  • 2023–present Meaning Alignment Institute Co-founder & Principal Investigator Lead research on thick models of value, moral graph elicitation, full-stack alignment, and post-market and democratic mechanisms. Principal investigator on the Full-Stack Alignment research program, a network of 50+ researchers at universities and industry labs. Collaborators include Tan Zhi-Xuan, Matija Franklin, Iason Gabriel, Jakob Foerster, Oliver Klingefjord, Ryan Lowe, Seth Lazar, etc.

Publications

  • Co-Aligning AI and Institutions with Thick Models of Value (2025). With Tan Zhi-Xuan, Matija Franklin, Michiel Bakker, Jakob Foerster, Iason Gabriel, et al. ICML 2025 Workshop on Models of Human Feedback for AI Alignment (MoFA), Vancouver. arXiv:2512.03399
  • What are Human Values, and How Do We Align AI to Them? (2024). With Oliver Klingefjord and Ryan Lowe. Funded by OpenAI. arXiv:2404.10636

In preparation: manuscripts in metaethics and action theory; political theory and institutional design; and mechanism design.

Previous Positions

  • 2007–2009 CouchSurfing International Reputation Systems & Metrics Joined when the site had ~100k active users. Developed the meaning-based organizational metrics that became the north star for the company and the primary criterion for feature rollouts, tracking tens of millions of in-person hosting experiences. Built the reputation and trust system, including ML models for user safety. [metrics writeup]

  • 2013–2015 Center for Humane Technology Co-founder Co-founded CHT with Tristan Harris. Led research on the design of healthier social-media ecosystems. Coined “Time Well Spent,” a framing and family of metrics subsequently adopted by teams at Facebook, Google, and Apple. Published mockups and designs for user-protection interventions that closely prefigured what later shipped as iOS Screen Time and Downtime. Briefed policymakers on political polarization, social isolation, declining trust, and internet addiction. Departed when the research track was moved outside CHT's scope.

  • 2015–2020 Independent Researcher Funded by Patrick Collison and Stripe. Delivered the influential talk “Is Anything Worth Maximizing?”, which led to recommender system changes at Facebook, Apple, and Google. Developed the philosophical foundations of new business metrics and design methods.

  • 2017–2022 School for Social Design Founder Online training for humane recommender systems. Trained product, metrics, and UX practitioners from Google, Facebook, Twitter, Notion, and other companies in values-based data science.

Workshops Organized

  • AGI-Ready Institution Design, Paris, February 2026. Co-located with IASEAI. Convened ~25 researchers from DeepMind, Oxford, UC Berkeley, ANU, GovAI, Mila, the Cooperative AI Foundation, and others across mechanism design, social choice, legal theory, political philosophy, and AI governance.
  • Thick Models of Choice, Oxford HAI Lab, March 2025. Co-organized with Philipp Koralus (Director, Human-Centered AI Lab, Oxford Institute for Ethics in AI). Significant portions of the Full-Stack Alignment paper (arXiv:2512.03399) emerged from this workshop.

Talks & Essays

  • Full-Stack Alignment, Post-AGI Civilizational Equilibria Workshop (2025)
  • Practical Techniques to Align Markets with Meaning (2023). [Video] · [Transcript]
  • Values-Based Data-Science and Design (2021). Textbook. textbook.sfsd.io
  • Social Programming Considered as a Habitat for Groups (2019). [PDF]
  • Nothing to be Done (2017). [Medium]
  • Is Anything Worth Maximizing? (2016). Cited as a direct influence on recommender-system changes at Facebook, Apple, and Google. [Video] · [Transcript]
  • Choicemaking and the Interface (2014). nxhx.org/Choicemaking

Selected interviews: Jim Rutt Show (EP 325: Full-Stack AI Alignment, 2025; Currents 080: Rebuilding Meaning with Ellie Hain, 2023); RadicalxChange Podcast with Matt Prewitt (2024); Clearer Thinking Podcast with Spencer Greenberg (2024).

Prototypes & Systems

  • Moral Graph Elicitation (2023). A democratic process for eliciting values from large, diverse groups and building a consensus graph over which values are wiser than which. Funded by OpenAI. [Writeup]

Education

Dartmouth College, A.B. Computer Science & Modeling, 2003 (enrolled 1999) Studied participatory music with Christian Wolff and playwriting with Peter Parnell.

MIT High School Studies Program Took MIT coursework for credit (later applied to Dartmouth degree); mentored by Howie Shrobe and Marvin Minsky.

Last updated: April 2026