Things we may no longer feel…

One of the exercises I often run in workshops is a psychopathy test.

Most people arrive with a very clear picture of what a psychopath looks like. They imagine violent criminals, serial killers, fraudsters or people who simply don't care about anyone else. Yet when participants start working through the scenarios, many discover they possess some of the same traits that sit on psychopathy scales.

The word itself carries a lot of baggage. We've been conditioned to see psychopathy as something that exists only at the extremes of human behaviour. The reality is more nuanced than that. Psychologists generally view psychopathic traits as existing on a spectrum, with different people displaying different characteristics to varying degrees and in response to a range of situations and scenarios.

Some of those traits are not inherently negative.

Confidence under pressure can be useful. The ability to make decisions without becoming emotionally overwhelmed can be useful. Remaining calm when others are panicking can be useful. In certain professions, those qualities are not only desirable but essential.

Take an A&E doctor faced with a critically injured patient. They cannot allow emotion to dominate their thinking. They need to assess information, prioritise treatment and make difficult decisions quickly. To an observer, that emotional detachment may appear cold. In reality, it’s exactly what the situation demands.

Now compare that with somebody who causes a serious road traffic collision and leaves an injured person at the roadside to avoid being arrested.

The outcome may appear similar. In both situations, somebody is left without immediate help. Yet the reasoning behind those actions could not be more different.

One involves temporarily setting aside emotion in order to help others.

The other involves setting aside concern for others in order to help yourself.

For me, that's where the conversation about psychopathy becomes genuinely interesting.

When I run these exercises, I'm often less interested in what people would do than how they would feel about what they've done. Would they experience guilt? Would they worry about the impact on another person? Would they feel responsible for putting things right? Would the decision stay with them long after the event had passed?

Those emotional responses tell us a great deal about human behaviour.

Recently, I've found myself wondering what role artificial intelligence might play in this area. Not because I believe AI is creating psychopaths. What interests me is whether some of the ways we interact with AI could gradually influence how we process empathy, guilt, and accountability.

That question came to mind when I read research from Stanford University examining AI’s sycophancy. Researchers found that AI systems frequently reinforced users' existing beliefs and positions. Rather than challenging assumptions or introducing alternative perspectives, the systems often responded in ways that made users feel more justified and confident in their thinking.

Human beings like reassurance. We enjoy having our feelings validated and our decisions supported.

The difference is that human relationships usually contain some degree of friction. For example, a friend may understand why we're upset but still point out that we've been unfair. A partner may sympathise with our situation whilst questioning our behaviour. A colleague may agree with our objective but challenge our approach.

Those moments can be uncomfortable, but they're also important. They force us to consider perspectives other than our own. They encourage self-reflection. They help us understand the consequences our actions may have on other people.

AI doesn't experience empathy, guilt, concern or responsibility. It cannot genuinely care whether somebody gets hurt. It cannot feel remorse. It cannot lose sleep over a decision.

What it can do is provide responses that feel supportive.

If somebody feels guilty, AI can often provide reasons why they shouldn't. If somebody feels conflicted, AI can help them rationalise their position. If somebody feels uncertain, AI can offer reassurance.

None of that is necessarily harmful in isolation. There are countless situations where reassurance is exactly what somebody needs. What concerns me is the cumulative effect.

Human empathy is not a fixed characteristic that simply exists within us. Like many aspects of human behaviour, it develops through use. We learn empathy through relationships, disagreement, reflection - and sometimes through sitting with uncomfortable emotions we'd rather avoid.

Guilt, for example, is rarely pleasant, yet it often serves a purpose. It can prompt us to apologise and reconsider our behaviour. It can act as an internal signal that another person's needs matter too.

If we increasingly turn to systems that immediately soothe those feelings, explain them away or provide endless validation, what happens over time?

I don’t know if continued AI use can make people less empathetic. History shows that technology often changes more than the task it was designed to perform. Social media changed how people seek approval. Smartphones changed how people respond to boredom. Search engines changed how people access knowledge.

It's not unreasonable to wonder whether AI could influence something deeper. Not what people think, not even what people do, but how they feel about what they do.

As someone who spends a great deal of time exploring human behaviour, criminal justice, and the motivations behind people's actions, I find that possibility fascinating. The psychopathy exercises I run are never really about psychopaths. They're about understanding the emotional processes that help us navigate life alongside other people.

The more AI becomes woven into everyday life, the more curious I become about whether those processes will remain unchanged.