This page is still being polished. If you have thoughts, please share them via the feedback form.
Data on this page is preliminary and may change. Please do not share or cite these figures publicly.
Foundational safety research, theoretical understanding, and scientific inquiry informing AI development.
Also in Engineering & Development
Methods to gain a comprehensive understanding of how large language models learn, reason, and scale, such as by examining in-context learning (ICL) mechanisms, the influence of data and design on behavior, the theoretical foundations of scaling, the emergence of advanced capabilities, and the nature of reasoning.
Reasoning
Investigates foundational model learning, reasoning, and scaling mechanisms through research.
Mechanistic understanding of In-Context Learning
Investigating the internal processes by which transformers perform ICL, including whether these processes resemble emergent optimization behavior, advanced pattern-matching, or other structural mechanisms. This research may include scenario-based analyses to identify the circuits critical for ICL under artificial constraints.
2.4.1 Research & FoundationsInfluences on ICL behavior and performance
Examining how the tasks, instructions, pre-training data distribution, and design choices (e.g., instruction tuning, model size, training duration) shape the range and reliability of behaviors that can be specified in-context.
2.4.1 Research & FoundationsTheoretical and representational aspects of scaling
Clarifying when and how scaling drives improvements, such as by building a more robust theoretical framework to describe scaling laws, or analyzing how increasing model size and training data influence learned representations.
2.4.1 Research & FoundationsEmergence and task-specific scaling patterns
Formalizing and forecasting the emergence of new capabilities as models scale, investigating whether scaling alone can produce certain capabilities, and designing methods for discovering task-specific scaling laws.
2.4.1 Research & FoundationsImpact of scaling and training on reasoning capabilities
Determining whether and how increases in model size and training complexity enhance reasoning abilities, and identifying which aspects of training conditions and data sources facilitate the acquisition of reasoning skills.
2.4.1 Research & FoundationsMechanistic understanding and limits of LLM reasoning
Examining the underlying mechanisms of reasoning in LLMs, exploring non-deductive reasoning capabilities of LLMs (e.g., causal or social reasoning).
2.4.1 Research & FoundationsLimits of Transformers
Defining the computational limits of transformers in supporting sophisticated reasoning.
2.4.1 Research & FoundationsTheoretical foundations and provable safety in AI systems
Advancing the theoretical foundations of AI safety by building models and frameworks that ensure provably correct and robust behavior. These efforts span from verifiable architectures and formal verification methods to embedded agency, decision theory, incentive structures aligned with causal reasoning, and control theory.
2.4.1 Research & FoundationsTheoretical foundations and provable safety in AI systems > Building verifiable and robust AI architectures
Constructing AI systems with architectures that support formal verification and robustness guarantees, such as world models that enable safe and reliable planning, or guaranteed safe AI with Bayesian oracles. This area emphasizes simplicity and transparency to aid in provability.
1.1.4 Model ArchitectureTheoretical foundations and provable safety in AI systems > Formal verification of AI systems
Applying formal methods to verify that AI models and algorithms meet stringent safety, robustness, and performance criteria. This includes proving resilience against adversarial inputs and perturbations, and certifying conformance to specified safety properties under varying conditions.
2.2.2 Testing & EvaluationTheoretical foundations and provable safety in AI systems > Decision theory and rational agency
Establishing formal decision-making frameworks that ensure rational and safe choices by AI agents, potentially drawing on concepts like causal and evidential decision theory.
2.4.1 Research & FoundationsTheoretical foundations and provable safety in AI systems > Embedded agency
Explores how agents can model and reason about themselves and their environment as interconnected parts of a single system, addressing challenges like self-reference, resource constraints, and the stability of reasoning processes. This includes tackling problems arising from the lack of a clear boundary between the agent and its environment.
2.4.1 Research & FoundationsTheoretical foundations and provable safety in AI systems > Causal incentives
Developing frameworks that formalize how to align agent incentives with safe and desired outcomes by ensuring their causal understanding matches intended objectives. This research provides a formal language for guaranteeing safety, addressing challenges like goal misspecification, and complementing broader efforts in agent foundations and robust system design.
2.4.1 Research & FoundationsExpert Survey: AI Reliability & Security Research Priorities
O'Brien, Joe; Dolan, Jeremy; Kim, Jay; Dykhuizen, Jonah; Sania, Jeba; Becker, Sebastian; Kraprayoon, Jam; Labrador, Cara (2025)
Our survey of 53 specialists across 105 AI reliability and security research areas identifies the most promising research prospects to guide strategic AI R&D investment. As companies are seeking to develop AI systems with broadly human-level capabilities, research on reliability and security is urgently needed to ensure AI's benefits can be safely and broadly realized and prevent severe harms. This study is the first to quantify expert priorities across a comprehensive taxonomy of AI safety and security research directions and to produce a data-driven ranking of their potential impact. These rankings may support evidence-based decisions about how to effectively deploy resources toward AI reliability and security research.
Plan and Design
Designing the AI system, defining requirements, and planning development
Developer
Entity that creates, trains, or modifies the AI system
Map
Identifying and documenting AI risks, contexts, and impacts
