Xavier Roberts-Gaal

Patricia Paskov, Kevin Wei, Shen Zhou Hong et al. · cmu

Human uplift studies - or studies that measure AI effects on human performance relative to a status quo, typically using randomized controlled trial (RCT) methodology - are increasingly used to inform deployment, governance, and safety decisions for frontier AI systems. While the methods underlying these studies are well-established, their interaction with the distinctive properties of frontier AI systems remains underexamined, particularly when results are used to inform high-stakes decisions. We present findings from interviews with 16 expert practitioners with experience conducting human uplift studies in domains including biosecurity, cybersecurity, education, and labor. Across interviews, experts described a recurring tension between standard causal inference assumptions and the object of study itself. Rapidly evolving AI systems, shifting baselines, heterogeneous and changing user proficiency, and porous real-world settings strain assumptions underlying internal, external, and construct validity, complicating the interpretation and appropriate use of uplift evidence. We synthesize these challenges across key stages of the human uplift research lifecycle and map them to practitioner-reported solutions, clarifying both the limits and the appropriate uses of evidence from human uplift studies in high-stakes decision-making.

Kevin L. Wei, Patricia Paskov, Sunishchal Dev et al.

In this position paper, we argue that human baselines in foundation model evaluations must be more rigorous and more transparent to enable meaningful comparisons of human vs. AI performance, and we provide recommendations and a reporting checklist towards this end. Human performance baselines are vital for the machine learning community, downstream users, and policymakers to interpret AI evaluations. Models are often claimed to achieve "super-human" performance, but existing baselining methods are neither sufficiently rigorous nor sufficiently well-documented to robustly measure and assess performance differences. Based on a meta-review of the measurement theory and AI evaluation literatures, we derive a framework with recommendations for designing, executing, and reporting human baselines. We synthesize our recommendations into a checklist that we use to systematically review 115 human baselines (studies) in foundation model evaluations and thus identify shortcomings in existing baselining methods; our checklist can also assist researchers in conducting human baselines and reporting results. We hope our work can advance more rigorous AI evaluation practices that can better serve both the research community and policymakers. Data is available at: https://github.com/kevinlwei/human-baselines

Shen Zhou Hong, Alex Kleinman, Alyssa Mathiowetz et al.

Large language models (LLMs) perform strongly on biological benchmarks, raising concerns that they may help novice actors acquire dual-use laboratory skills. Yet, whether this translates to improved human performance in the physical laboratory remains unclear. To address this, we conducted a pre-registered, investigator-blinded, randomized controlled trial (June-August 2025; n = 153) evaluating whether LLMs improve novice performance in tasks that collectively model a viral reverse genetics workflow. We observed no significant difference in the primary endpoint of workflow completion (5.2% LLM vs. 6.6% Internet; P = 0.759), nor in the success rate of individual tasks. However, the LLM arm had numerically higher success rates in four of the five tasks, most notably for the cell culture task (68.8% LLM vs. 55.3% Internet; P = 0.059). Post-hoc Bayesian modeling of pooled data estimates an approximate 1.4-fold increase (95% CrI 0.74-2.62) in success for a "typical" reverse genetics task under LLM assistance. Ordinal regression modelling suggests that participants in the LLM arm were more likely to progress through intermediate steps across all tasks (posterior probability of a positive effect: 81%-96%). Overall, mid-2025 LLMs did not substantially increase novice completion of complex laboratory procedures but were associated with a modest performance benefit. These results reveal a gap between in silico benchmarks and real-world utility, underscoring the need for physical-world validation of AI biosecurity assessments as model capabilities and user proficiency evolve.

Carson Ezell, Xavier Roberts-Gaal, Alan Chan

As AI agents become more widely deployed, we are likely to see an increasing number of incidents: events involving AI agent use that directly or indirectly cause harm. For example, agents could be prompt-injected to exfiltrate private information or make unauthorized purchases. Structured information about such incidents (e.g., user prompts) can help us understand their causes and prevent future occurrences. However, existing incident reporting processes are not sufficient for understanding agent incidents. In particular, such processes are largely based on publicly available data, which excludes useful, but potentially sensitive, information such as an agent's chain of thought or browser history. To inform the development of new, emerging incident reporting processes, we propose an incident analysis framework for agents. Drawing on systems safety approaches, our framework proposes three types of factors that can cause incidents: system-related (e.g., CBRN training data), contextual (e.g., prompt injections), and cognitive (e.g., misunderstanding a user request). We also identify specific information that could help clarify which factors are relevant to a given incident: activity logs, system documentation and access, and information about the tools an agent uses. We provide recommendations for 1) what information incident reports should include and 2) what information developers and deployers should retain and make available to incident investigators upon request. As we transition to a world with more agents, understanding agent incidents will become increasingly crucial for managing risks.

Neil Natarajan, Sruthi Viswanathan, Xavier Roberts-Gaal et al.

Static solutions don't serve a dynamic mind. Thus, we advocate a shift from static mental health diagnostic assessments to continuous, artificial intelligence (AI)-driven assessment. Focusing on Attention-Deficit/Hyperactivity Disorder (ADHD) as a case study, we explore how generative AI has the potential to address current capacity constraints in neuropsychology, potentially enabling more personalized and longitudinal care pathways. In particular, AI can efficiently conduct frequent, low-level experience sampling from patients and facilitate diagnostic reconciliation across care pathways. We envision a future where mental health care benefits from continuous, rich, and patient-centered data sampling to dynamically adapt to individual patient needs and evolving conditions, thereby improving both accessibility and efficacy of treatment. We further propose the use of mental health digital twins (MHDTs) - continuously updated computational models that capture individual symptom dynamics and trajectories - as a transformative framework for personalized mental health care. We ground this framework in empirical evidence and map out the research agenda required to refine and operationalize it.