Psychology for Human–AI Systems
Psyemic brings psychological science to the safety, performance and wellbeing of human–AI systems.
As AI systems become more capable and more embedded in decision-making, their safety and performance is shaped by the interaction between technical design and human behaviour.
Psyemic studies these interactions and helps organisations design, test and deploy human–AI systems that remain safe, effective and reliable in real-world use.
Why Psychology Matters
The most important risks and opportunities in AI do not arise from the model alone. They emerge through interaction between technical systems and human behaviour.
How people understand AI systems, when they trust them, when they challenge them, and how they use them under pressure can shape outcomes as much as the underlying technology.
A system that performs well in testing may still create risks in practice if users over-trust it, defer judgment, or rely on it in unintended ways.
These behavioural dynamics need to be examined directly through realistic testing of human–AI systems.
These dynamics are particularly important in systems that interact directly with users, where issues such as over-reliance, influence, or vulnerability may shape outcomes over time.
Psyemic brings a clinical psychology perspective to this work. Clinical psychology is a discipline built around understanding behaviour, judgment and risk in complex, high-stakes environments. We apply that perspective to AI safety and deployment.
About Us
Psyemic takes its name from the idea of an emic perspective in psychology:
Understanding behaviour from the inside, in context.
We are Consultant Clinical Psychologists with extensive post-qualification experience following training on the NHS–UCL Doctoral Programme. Our backgrounds were developed in high-stakes healthcare and government settings, where understanding behaviour under uncertainty and management of risk is essential.
Clinical psychology trains practitioners to assess complex behaviour, manage risk and support sound decision-making in difficult conditions. We now apply that expertise to human–AI systems.
Our early work explored the psychological and societal risks of conversational AI. Since then, we have contributed to multidisciplinary AI safety and red-teaming projects focused on the behavioural dynamics of advanced systems.
Psyemic brings applied psychological expertise to organisations developing and deploying AI.
Our Services
Organisations work with Psyemic when they need to understand what is happening once people begin interacting with AI systems — and what this means for safety, performance and user wellbeing.
Our work focuses on the psychological and behavioural dynamics through which safety, performance and impact on users emerge.
We typically support organisations in the following ways:
-
We support behavioural and psychological red-teaming of AI systems, complementing technical approaches by examining how human behaviour shapes risk.
This includes developing psychologically realistic user personas to evaluate how different types of users interact with AI systems, including user vulnerabilities and situations where trust, judgment or decision-making may be affected.
This also includes examining how users may attempt to manipulate or bypass system safeguards, and how models respond under adversarial or boundary-testing conditions.
-
We help organisations design and deploy AI systems that enhance human performance without eroding judgment, responsibility or skill.
Our focus is on human–AI systems that support effective decision-making in complex, operational settings.
-
We help organisations define how AI systems should behave in practice, particularly in high-stakes, sensitive or complex environments.
This includes identifying unsafe or counterproductive interaction patterns and shaping design and policy approaches that support more reliable and context-appropriate behaviour.
-
AI systems increasingly shape how people think, learn and work.
We help organisations understand these effects and develop approaches that support safe, responsible and sustainable deployment.
The Team
-
Dr Rick Hennessy
CO-FOUNDER
Rick is a Consultant Clinical Psychologist and Senior Visiting Fellow at The Alan Turing Institute’s Centre for Emerging Technology and Security (CETaS).
His work focuses on the psychological aspects of human-AI safety, security and teaming across government and industry. He is particularly interested in how psychological insight can strengthen human–AI systems, improve decision-making and support responsible AI adoption.
His wider experience spans mental health, clinical neuropsychology, insider risk, behaviour change, research, innovation and leadership across senior roles in the NHS, government, academia and the private sector.
He co-founded Psyemic to help ensure that psychological expertise plays a central role in the safe development and deployment of advanced AI systems.
-
Dr Tim Kember
CO-FOUNDER
Tim is a Consultant Clinical Psychologist specialising in the application of psychological science to complex, high-stakes problems.
Throughout his career, he has focused on areas where psychological expertise is most needed and least represented. His work now sits at the intersection of human behaviour, risk and advanced technology.
He specialises in the psychological aspects of AI safety and the human factors involved in AI deployment, helping organisations understand and manage the behavioural risks introduced by advanced AI systems.
He co-founded Psyemic to bring rigorous psychological thinking to the emerging challenges of human–AI systems.
Get In touch
Developing or deploying advanced AI systems means thinking not only about what the model can do, but how people will actually use it in practice.
Psyemic works with organisations to understand and address the behavioural dynamics that shape safety, performance and user impact in real-world human–AI systems.
If you are:
red teaming AI models or applications
preparing an AI system for deployment
exploring behavioural or user safety risks
introducing AI into high-stakes or sensitive environments
trying to understand how people are actually using your systems
we would be glad to speak.