![[Ram Rachum.jpg|right40]] I'm Ram Rachum, and this is the knowledge base for my research. It has many notes of thoughts that I have that I hope will lead me to interesting results. Feel free to dive in. My mission statement: >**I want to ensure that Artificial Intelligence will benefit human society rather than hurt it. I'm addressing both short-term consequences of existing AI tools such as LLMs and long-term existential risk imposed by the potential development of Artificial General Intelligence.** > >**I propose to solve the problem of AI Safety by improving the social behavior of AI agents. Most AI research is focused on getting AI agents to solve bigger and more complex problems in the most efficient way possible; my research is focused on achieving cooperation, [[Emergent reciprocity|reciprocity]] and teamwork between AI agents.** > >**My strategy is to distill lessons from the study of animal and human societies and apply them to AI agents.** In other words, I'm applying [Multi-Agent Reinforcement Learning](https://en.wikipedia.org/wiki/Multi-agent_reinforcement_learning) to [AI Safety](https://en.wikipedia.org/wiki/AI_safety). My approach is within the schools of thought of [Cooperative AI](https://www.nature.com/articles/d41586-021-01170-0) and [Multipolar AI](https://nickbostrom.com/papers/openness.pdf). I'm affiliated with [the GOLD lab](https://sites.google.com/site/dekelreuth/gold-lab) at Tufts University, led by [Professor Reuth Mirsky](https://sites.google.com/site/dekelreuth/welcome). I've received funding / fellowships / internships from: * [ALTER](https://alter.org.il/) * [Center for Human-Compatible Artificial Intelligence (CHAI)](https://humancompatible.ai/) * [The Future of Life Institute (FLI)](https://futureoflife.org) * [Nonlinear](https://www.nonlinear.org/) I'll be happy to get any comments and feedback. Email me at [email protected] [[About Ram Rachum|More about me]]. ## 📜 My papers * [[Dominance hierarchies|Emergent Dominance Hierarchies in Reinforcement Learning Agents]] (2024) * [[Stubborn|Stubborn: An Environment for Evaluating Stubbornness between Agents with Aligned Incentives]] (2023) * Poster: [Using Sequential Social Dilemmas to Produce Emergent Aligned AI](https://r.rachum.com/aisic2022-poster) (2022, won outstanding poster at AISIC 2022) * [[Fruit Slots|Fruit Slots: Experimenting with an Autocurriculum of Implicit Communication in Reinforcement Learning Agents]] (2022) ## 💬 My posts * [Can AI agents learn to be good?](https://futureoflife.org/ai-research/can-ai-agents-learn-to-be-good/) My guest post on FLI's blog. * A Conservative Vision for AI Alignment, a LessWrong sequence with David Manheim: * Part 1: [A Conservative Vision for AI Alignment](https://r.rachum.com/conservative-alignment-1) * Part 2: [Messy on Purpose](https://r.rachum.com/conservative-alignment-2) * Part 3: [12 Angry Agents, or: A Plan for AI Empathy](https://r.rachum.com/conservative-alignment-3) ## ⭐ Sign up to get updates Sign up to my [research mailing list](http://r.rachum.com/announce) to get monthly updates about my research. Every month I outline the goals for that month, and evaluate my progress on last month's goals.