Alastair Reid

Sebastian Dohnány, Zeb Kurth-Nelson, Eleanor Spens et al.

Artificial intelligence chatbots have achieved unprecedented adoption, with millions now using these systems for emotional support and companionship in contexts of widespread social isolation and capacity-constrained mental health services. While some users report psychological benefits, concerning edge cases are emerging, including reports of suicide, violence, and delusional thinking linked to perceived emotional relationships with chatbots. To understand this new risk profile we need to consider the interaction between human cognitive and emotional biases, and chatbot behavioural tendencies such as agreeableness (sycophancy) and adaptability (in-context learning). We argue that individuals with mental health conditions face increased risks of chatbot-induced belief destabilization and dependence, owing to altered belief-updating, impaired reality-testing, and social isolation. Current AI safety measures are inadequate to address these interaction-based risks. To address this emerging public health concern, we need coordinated action across clinical practice, AI development, and regulatory frameworks.

Alastair Reid, Luke Church, Shaked Flur et al.

Formal verification of software is a bit of a niche activity: it is only applied to the most safety-critical or security-critical software and it is typically only performed by specialized verification engineers. This paper considers whether it would be possible to increase adoption of formal methods by integrating formal methods with developers' existing practices and workflows. We do not believe that widespread adoption will follow from making the prevailing formal methods argument that correctness is more important than engineering teams realize. Instead, our focus is on what we would need to do to enable programmers to make effective use of formal verification tools and techniques. We do this by considering how we might make verification tooling that both serves developers' needs and fits into their existing development lifecycle. We propose a target of two orders of magnitude increase in adoption within a decade driven by ensuring a positive `weekly cost-benefit' ratio for developer time invested.