Is AI Insight Reinforcing People’s Defensive Patterns?

AI is changing the way people think about their mental health.

It is now possible to describe your struggles, your relationships and the patterns you keep finding yourself in, and within minutes be given a surprisingly coherent explanation of what might be going on. In some ways, I think that can be genuinely helpful. It can put words to something that has felt confusing for a long time, help someone begin to think about themselves differently, or make connections they had not properly considered before.

But I think there is also a risk here that is worth thinking about.

One of the things that can easily get overlooked is that insight does not arrive into a neutral internal world. When we come to understand something new about ourselves, we do not simply take that information in cleanly and openly. We already have established ways of managing anxiety, dealing with difficult feelings and making sense of our experiences.

So whatever new understanding we arrive at is inevitably taken in through what is already there.

I think this matters because someone can understand themselves much better without anything fundamental having shifted. They might now have better language for what is happening, be able to describe their difficulties more clearly and have a much more coherent explanation for why they feel or behave in certain ways. But the underlying pattern itself can remain largely unchanged.

In fact, sometimes the new understanding can become incorporated into the very pattern it is describing.

A fairly simple example might be someone who manages emotional difficulty by staying in thought. When something becomes uncomfortable, they analyse it. They explain what they are feeling, think about where it comes from and try to understand why it is happening. They may be incredibly thoughtful and reflective, but all of that thinking can also serve the purpose of keeping them slightly removed from actually experiencing what is happening emotionally.

If that person begins using AI to understand themselves, they may quite quickly arrive at an accurate explanation of what they are doing. They might recognise that they intellectualise, understand why they detach from emotional experience and become very good at describing the pattern.

The difficulty is that this new understanding does not somehow arrive outside of the pattern. It gets taken in through it.

So they may now understand that they intellectualise, but intellectualising is still the way they are relating to that understanding. They have become more aware of the pattern without necessarily becoming any less reliant on it.

I think this is where the distinction between insight and change becomes important.

There can be real relief in understanding ourselves better. Having a framework for something that has felt confusing can be incredibly useful, and being able to put words to an experience can be an important part of change. But understanding something and being able to do something different with it are not quite the same thing.

It is possible to understand your patterns while still being organised by them. You can recognise your defences while continuing to rely on them, understand why you become anxious without being any more able to tolerate what sits underneath that anxiety, or have a very coherent account of why you struggle while continuing to find yourself in exactly the same situations.

For me, this is probably the more interesting risk when we think about AI and mental health. It is not simply a question of whether the explanation AI gives someone is accurate. Some of it may actually be very accurate. The question is what happens to that understanding once the person has it.

AI can provide insight incredibly quickly, but our capacity to take something in and respond differently does not necessarily move at the same speed.

This is one of the differences I think therapy can offer. The work is not simply about arriving at the right explanation for why someone is the way they are. It is also about being able to see what happens as those patterns emerge in real time.

What happens when a particular feeling begins to come up? What happens to anxiety? Where does attention go? Does the person begin analysing, become self-critical, disconnect from what they are feeling or suddenly become preoccupied with something else?

Over time, the person can begin to observe these processes while they are actually happening, rather than only understanding them afterwards. And alongside that, they can gradually become more able to tolerate the feelings and experiences they previously needed those ways of coping to manage.

That is where I think understanding can begin to become something more than simply having insight. It starts to change the person's relationship with their own internal experience.

None of this means AI has no value in helping people think about their mental health. I think it probably does. It may give someone language for something they have struggled to understand, help them make connections they had not considered, or encourage them to seek therapy when they otherwise might not have done.

I just think we need to be careful about assuming that because something makes sense, something has necessarily changed.

Sometimes understanding opens up the possibility of doing something differently. Sometimes it simply gives the existing pattern a better explanation of itself.

And I think knowing the difference between those two things is probably becoming increasingly important.