Over the past year, veteran software engineer Jay Prakash Thakur has spent his nights and weekends prototyping AI agents that could, in the near future, order meals and engineer mobile apps almost entirely on their own. His agents, while surprisingly capable, have also exposed new legal questions that await companies trying to capitalize on Silicon Valley's hottest new technology. Agents are AI programs that can act mostly independently, allowing companies to automate tasks such as answering customer questions or paying invoices. While ChatGPT and similar chatbots can draft emails or analyze bills upon request, Microsoft and other tech giants expect that agents will tackle more complex functions--and most importantly, do it with little human oversight. The tech industry's most ambitious plans involve multi-agent systems, with dozens of agents someday teaming up to replace entire workforces.

During the 33rd International Joint Conference on Artificial Intelligence (IJCAI), held in Jeju, I had the opportunity to meet with one of the keynote speakers, Gillian Hadfield. We spoke about her interdisciplinary research, career trajectory, path into AI alignment, law, and general thoughts on AI systems. Transcript: Note: the transcript has been lightly edited for clarity. This is an interview with Professor Gillian Hadfield who was a keynote speaker at IJCAI 2024. She gave a very insightful talk about normative infrastructures and how they can guide our search for AI alignment. Kumar Kshitij Patel (KKP): Could you talk a bit about your background and career trajectory? I want our readers to understand how much interdisciplinary work you've done over the years. Gillian Hadfield (GH): I did a PhD in economics and a law degree, a JD, at Stanford, originally motivated by wanting to think about the big questions about the world. So I read John Rawls' theory of justice when I was an undergraduate, and those are the big questions: how do we organize the world and just institutions, but I was very interested in using more formal methods and social scientific approaches. That's why I decided to do that joint degree. So, this is in the 1980s, and in the early days of starting to use a lot of game theory. I studied information theory, a student of Canaro and Paul Milgram at the economics department at Stanford. I did work on contract theory, bargaining theory, but I was still very interested in going to law school, not to practice law, but to learn about legal institutions and how those work. I was a member of this emerging area of law and economics early in my career, which of course, was interdisciplinary, using economics to think about law and legal institutions.

B.4 Complete set of experimental results associated to section 4 In this section we display the complete set of results associated to figures shown in section 4. We display in figure 2 the rewards of all agents during training (calibrator, merchant on supertype 1 and n 1 merchants on supertype 2) for experiments 1-5 previously described.

The communication rules used to solve each benchmark are not chosen manually.

Large Language Models (LLMs) trained on massive corpora have shown remarkable success in knowledge-intensive tasks. Yet, most of them rely on pre-stored knowledge. Inducing new general knowledge from a specific environment and performing reasoning with the acquired knowledge--situated inductive reasoning, is crucial and challenging for machine intelligence. In this paper, we design Mars, an interactive environment devised for situated inductive reasoning. It introduces counter-commonsense game mechanisms by modifying terrain, survival setting and task dependency while adhering to certain principles.

The race to deploy AI agents is heating up. At its annual I/O developer conference yesterday, Google announced that Jules, its new AI coding assistant, is now available worldwide in public beta. The launch marks the company's latest effort to corner the burgeoning market for AI agents, widely regarded across Silicon Valley as essentially a more practical and profitable form of chatbot. Virtually every other major tech giant -- including Meta, OpenAI, and Amazon, just to name a few -- has launched its own agent product in recent months. Also: I tested ChatGPT's Deep Research against Gemini, Perplexity, and Grok AI to see which is best Originally unveiled by Google Labs in December, Jules is positioned as a reliable, automated coding assistant that can manage a broad suite of time-consuming tasks on behalf of human users. The model is "asynchronous," which, in programming-speak, means it can start and work on tasks without having to wait for any single one of them to finish.

At a company-wide hackathon this month, developers at finance firm Block built a dizzying number of prototype tools including a database debugger, a program for identifying duplicated code, and an app that automates Bitcoin support. The sudden productivity boost was driven by Goose, an artificial intelligence agent developed by Block several months ago that can help with coding and other work like knocking together data visualizations or mocking up new product features. "We've always had really strong hack weeks, but this one was at another level," says Jackie Brosamer, who leads the AI and data platform at Block. "We have tens of ideas that we're looking to bring to production." Goose helped developers at Block to develop a new agent-to-agent communication server at the hackathon.

Multi-agent control is a central theme in the Cyber-Physical Systems (CPS). However, current control methods either receive non-Markovian states due to insufficient sensing and decentralized design, or suffer from poor convergence. This paper presents the Delayed Propagation Transformer (DePT), a new transformerbased model that specializes in the global modeling of CPS while taking into account the immutable constraints from the physical world. DePT induces a cone-shaped spatial-temporal attention prior, which injects the information propagation and aggregation principles and enables a global view. With physical constraint inductive bias baked into its design, our DePT is ready to plug and play for a broad class of multi-agent systems. The experimental results on one of the most challenging CPS - network-scale traffic signal control system in the open world - show that our model outperformed the state-of-the-art expert methods on synthetic and real-world datasets.

We perform volume analysis of Multiplicative Weights Updates (MWU) and its optimistic variant (OMWU) in zero-sum as well as coordination games. Our analysis provides new insights into these game/dynamical systems, which seem hard to achieve via the classical techniques within Computer Science and ML. First, we examine these dynamics not in their original space (simplex of actions) but in a dual space (aggregate payoffs of actions). Second, we explore how the volume of a set of initial conditions evolves over time when it is pushed forward according to the algorithm. This is reminiscent of approaches in evolutionary game theory where replicator dynamics, the continuous-time analogue of MWU, is known to preserve volume in all games. Interestingly, when we examine discrete-time dynamics, the choices of the game and the algorithm both play a critical role. So whereas MWU expands volume in zero-sum games and is thus Lyapunov chaotic, we show that OMWU contracts volume, providing an alternative understanding for its known convergent behavior. Yet, we also prove a no-free-lunch type of theorem, in the sense that when examining coordination games the roles are reversed. Using these tools, we prove two novel, rather negative properties of MWU in zero-sum games.

Progress in Reinforcement Learning (RL) algorithms goes hand-in-hand with the development of challenging environments that test the limits of current methods. While existing RL environments are either sufficiently complex or based on fast simulation, they are rarely both. Here, we present the NetHack Learning Environment (NLE), a scalable, procedurally generated, stochastic, rich, and challenging environment for RL research based on the popular single-player terminalbased roguelike game, NetHack. We argue that NetHack is sufficiently complex to drive long-term research on problems such as exploration, planning, skill acquisition, and language-conditioned RL, while dramatically reducing the computational resources required to gather a large amount of experience. We compare NLE and its task suite to existing alternatives, and discuss why it is an ideal medium for testing the robustness and systematic generalization of RL agents. We demonstrate empirical success for early stages of the game using a distributed Deep RL baseline and Random Network Distillation exploration, alongside qualitative analysis of various agents trained in the environment.