Scalable oversight, the process by which weaker AI systems supervise stronger ones, has been proposed as a key strategy to control future superintelligent systems. However, it is still unclear how scalable oversight itself scales. To address this gap, we propose a framework that quantifies the probability of successful oversight as a function of the capabilities of the overseer and the system being overseen. Specifically, our framework models oversight as a game between capability-mismatched players; the players have oversight-specific Elo scores that are a piecewise-linear function of their general intelligence, with two plateaus corresponding to task incompetence and task saturation. We validate our framework with a modified version of the game Nim and then apply it to four oversight games: Mafia, Debate, Backdoor Code and Wargames. For each game, we find scaling laws that approximate how domain performance depends on general AI system capability. We then build on our findings in a theoretical study of Nested Scalable Oversight (NSO), a process in which trusted models oversee untrusted stronger models, which then become the trusted models in the next step. We identify conditions under which NSO succeeds and derive numerically (and in some cases analytically) the optimal number of oversight levels to maximize the probability of oversight success. We also apply our theory to our four oversight games, where we find that NSO success rates at a general Elo gap of 400 are 13.5% for Mafia, 51.7% for Debate, 10.0% for Backdoor Code, and 9.4% for Wargames; these rates decline further when overseeing stronger systems.

Wargames are simulations of conflicts in which participants' decisions influence future events. While casual wargaming can be used for entertainment or socialization, serious wargaming is used by experts to explore strategic implications of decision-making and experiential learning. In this paper, we take the position that Artificial Intelligence (AI) systems, such as Language Models (LMs), are rapidly approaching human-expert capability for strategic planning -- and will one day surpass it. Military organizations have begun using LMs to provide insights into the consequences of real-world decisions during _open-ended wargames_ which use natural language to convey actions and outcomes. We argue the ability for AI systems to influence large-scale decisions motivates additional research into the safety, interpretability, and explainability of AI in open-ended wargames. To demonstrate, we conduct a scoping literature review with a curated selection of 100 unclassified studies on AI in wargames, and construct a novel ontology of open-endedness using the creativity afforded to players, adjudicators, and the novelty provided to observers. Drawing from this body of work, we distill a set of practical recommendations and critical safety considerations for deploying AI in open-ended wargames across common domains. We conclude by presenting the community with a set of high-impact open research challenges for future work.

We propose Monte Carlo Permutation Search (MCPS), a general-purpose Monte Carlo Tree Search (MCTS) algorithm that improves upon the GRAVE algorithm. MCPS is relevant when deep reinforcement learning is not an option, or when the computing power available before play is not substantial, such as in General Game Playing, for example. The principle of MCPS is to include in the exploration term of a node the statistics on all the playouts that contain all the moves on the path from the root to the node. We extensively test MCPS on a variety of games: board games, wargame, investment game, video game and multi-player games. MCPS has better results than GRAVE in all the two-player games. It has equivalent results for multi-player games because these games are inherently balanced even when players have different strengths. We also show that using abstract codes for moves instead of exact codes can be beneficial to both MCPS and GRAVE, as they improve the permutation statistics and the AMAF statistics. We also provide a mathematical derivation of the formulas used for weighting the three sources of statistics. These formulas are an improvement on the GRAVE formula since they no longer use the bias hyperparameter of GRAVE. Moreover, MCPS is not sensitive to the ref hyperparameter.

There is an increasing interest in using language models (LMs) for automated decision-making, with multiple countries actively testing LMs to aid in military crisis decision-making. To scrutinize relying on LM decision-making in high-stakes settings, we examine the inconsistency of responses in a crisis simulation ("wargame"), similar to reported tests conducted by the US military. Prior work illustrated escalatory tendencies and varying levels of aggression among LMs but were constrained to simulations with pre-defined actions. This was due to the challenges associated with quantitatively measuring semantic differences and evaluating natural language decision-making without relying on pre-defined actions. In this work, we query LMs for free form responses and use a metric based on BERTScore to measure response inconsistency quantitatively. Leveraging the benefits of BERTScore, we show that the inconsistency metric is robust to linguistic variations that preserve semantic meaning in a question-answering setting across text lengths. We show that all five tested LMs exhibit levels of inconsistency that indicate semantic differences, even when adjusting the wargame setting, anonymizing involved conflict countries, or adjusting the sampling temperature parameter $T$. Further qualitative evaluation shows that models recommend courses of action that share few to no similarities. We also study the impact of different prompt sensitivity variations on inconsistency at temperature $T = 0$. We find that inconsistency due to semantically equivalent prompt variations can exceed response inconsistency from temperature sampling for most studied models across different levels of ablations. Given the high-stakes nature of military deployment, we recommend further consideration be taken before using LMs to inform military decisions or other cases of high-stakes decision-making.

Drawing on the experiences of facilitators who have overseen 43 games over a four-year period, we illuminate recurring patterns, strategies, and decision-making processes observed during gameplay. Our analysis reveals key strategic considerations about AI development trajectories in this simulated environment, including: the destabilising effects of AI races, the crucial role of international cooperation in mitigating catastrophic risks, the challenges of aligning corporate and national interests, and the potential for rapid, transformative change in AI capabilities. We highlight places where we believe the game has been effective in exposing participants to the complexities and uncertainties inherent in AI governance. Key recurring gameplay themes include the emergence of international agreements, challenges to the robustness of such agreements, the critical role of cybersecurity in AI development, and the potential for unexpected crises to dramatically alter AI trajectories. By documenting these insights, we aim to provide valuable foresight for policymakers, industry leaders, and researchers navigating the complex landscape of AI development and governance.

To some, the advent of artificial intelligence (AI) promises better decision-making and increased military effectiveness while reducing the influence of human error and emotions. However, there is still debate about how AI systems, especially large language models (LLMs), behave compared to humans in high-stakes military decision-making scenarios with the potential for increased risks towards escalation and unnecessary conflicts. To test this potential and scrutinize the use of LLMs for such purposes, we use a new wargame experiment with 107 national security experts designed to look at crisis escalation in a fictional US-China scenario and compare human players to LLM-simulated responses in separate simulations. Wargames have a long history in the development of military strategy and the response of nations to threats or attacks. Here, we show a considerable high-level agreement in the LLM and human responses and significant quantitative and qualitative differences in individual actions and strategic tendencies. These differences depend on intrinsic biases in LLMs regarding the appropriate level of violence following strategic instructions, the choice of LLM, and whether the LLMs are tasked to decide for a team of players directly or first to simulate dialog between players. When simulating the dialog, the discussions lack quality and maintain a farcical harmony. The LLM simulations cannot account for human player characteristics, showing no significant difference even for extreme traits, such as "pacifist" or "aggressive sociopath". Our results motivate policymakers to be cautious before granting autonomy or following AI-based strategy recommendations.

Wargames are a powerful tool for understanding and rehearsing real-world decision making. Automated play of wargames using artificial intelligence (AI) enables possibilities beyond those of human-conducted games, such as playing the game many times over to see a range of possible outcomes. There are two categories of wargames: quantitative games, with discrete types of moves, and qualitative games, which revolve around open-ended responses. Historically, automation efforts have focused on quantitative games, but large language models (LLMs) make it possible to automate qualitative wargames. We introduce "Snow Globe," an LLM-powered multi-agent system for playing qualitative wargames. With Snow Globe, every stage of a text-based qualitative wargame from scenario preparation to post-game analysis can be optionally carried out by AI, humans, or a combination thereof. We describe its software architecture conceptually and release an open-source implementation alongside this publication. As case studies, we simulate a tabletop exercise about an AI incident response and a political wargame about a geopolitical crisis. We discuss potential applications of the approach and how it fits into the broader wargaming ecosystem.

In this unprecedented era of technology-driven transformation, it becomes more critical than ever that we aggressively invest in developing robust artificial intelligence (AI) for wargaming in support of decision-making. By advancing AI-enabled systems and pairing these with human judgment, we will be able to enhance all-domain awareness, improve the speed and quality of our decision cycles, offer recommendations for novel courses of action, and more rapidly counter our adversary's actions. It therefore becomes imperative that we accelerate the development of AI to help us better address the complexity of modern challenges and dilemmas that currently requires human intelligence and, if possible, attempt to surpass human intelligence--not to replace humans, but to augment and better inform human decision-making at machine speed. Although deep reinforcement learning continues to show promising results in intelligent agent behavior development for the long-horizon, complex tasks typically found in combat modeling and simulation, further research is needed to enable the scaling of AI to deal with these intricate and expansive state-spaces characteristic of wargaming for either concept development, education, or analysis. To help address this challenge, in our research, we are developing and implementing a hierarchical reinforcement learning framework that includes a multi-model approach and dimension-invariant observation abstractions.

When I was a much younger person, there were a couple of popular movies offering warnings about the dangers of unchecked artificial intelligence. The first is 1983′s "WarGames," in which a young computer hacker played by Matthew Broderick accidentally triggers a countdown to the launch of the full arsenal of the United States nuclear stockpile at the Soviet Union because the Pentagon had handed control of the intercontinental ballistic missile system to a computer program, following the failure of humans to execute launch orders during a training exercise. The second one is 1984′s "The Terminator," where killer robot Arnold Schwarzenegger is dispatched back in time by the sentient artificial intelligence (called Skynet) in order to assassinate the hero of the resistance that is fighting the artificial intelligence in the future. I think I have that right. I honestly never understood the whole time travel aspect of the "Terminator" franchise.

On the second floor of the stately King's College London building on the Strand, Vladimir Putin, Emmanuel Macron, Olaf Scholz and Joe Biden are sitting around a table studying a map of Ukraine. They are here to negotiate the future of the country, but they all have ulterior objectives too. Germany wants to ensure the safe transit of refugees; the US wants Russia to cease its disinformation campaign; France wants trade; and Russia needs dozens of sanctions to be lifted. But nobody is giving anything away. It's tense as hell and the clock is ticking.