The Third Person in the Therapy Room May Be AI: How AI Is Changing Therapy

A client walks into therapy after an argument with their partner. But this time, the therapist isn't the first place they have taken the story.

The night before, while the argument was still fresh, the client opened an AI chatbot and typed out what happened. They described what their partner said, explained how they responded, added some history about the relationship, and asked a question that millions of people are increasingly comfortable asking technology: “Am I overreacting, or is this behaviour unhealthy?”

The AI responded immediately. It helped organise the situation, suggested possible communication patterns, mentioned boundaries, offered questions for reflection, and perhaps introduced psychological language the client had not previously considered. The client continued the conversation. They added more details. The AI responded again. By the time they went to bed, the event no longer existed in their mind simply as an uncomfortable argument. It had been explained, categorised, and given a narrative.

Then they arrived at therapy. The therapist was still speaking with one client. Yet in a subtle way, another voice had already entered the room.

This may become one of the more interesting changes facing mental healthcare professionals as artificial intelligence becomes part of everyday life. The concern is not simply that people are using AI for mental health information. Search engines, books, podcasts, friends, and social media have influenced how people understand themselves for years.

AI introduces something different. It doesn't merely provide information for someone to read. It participates in a conversation about the person's experience.

And that means therapists may increasingly meet clients whose stories have already been interpreted before therapy even begins.

 

We Have Always Brought Other People's Voices Into Therapy

The idea that clients arrive in therapy with outside interpretations isn't new.

Someone experiencing relationship difficulties may have already spoken to three friends before their appointment. A parent may have shared an opinion. A colleague may have offered advice. A podcast may have introduced the idea of attachment styles. A social media post may have made the person wonder whether a partner is emotionally unavailable. Google may have provided a list of symptoms that suddenly made several experiences feel connected.

Human beings rarely experience events without trying to explain them.

We ask other people what they think because another perspective helps us make sense of uncertainty. Sometimes those conversations are useful. Sometimes they reinforce what we already believe. Sometimes they make situations considerably more confusing.

Therapists have always worked with these influences.

What makes AI different is the combination of availability, responsiveness, and apparent personalisation.

A friend may become tired after hearing about the same relationship problem for the fifth time. AI doesn't. A family member may be unavailable at two in the morning. AI isn't. A book cannot respond when someone says, “But that isn't exactly what happened in my situation.” AI can.

The result is an entirely new kind of pre-therapy conversation.

Someone can spend an hour explaining an experience, correcting the response, adding context, asking follow-up questions, exploring alternative interpretations, and requesting increasingly specific advice.

By the time they reach therapy, they may not simply have an opinion about what happened. They may have an entire framework for understanding it.

 

The Story May Arrive Already Labelled

Consider the difference between these two statements.

"Whenever we argue, my partner stops talking to me for a few hours."

And:

"I think my partner uses stonewalling because they have an avoidant attachment style."

The second statement contains much more than an observation. It contains an interpretation.

Perhaps that interpretation is useful. Perhaps it is partly accurate. Perhaps it is completely wrong. The important point is that the therapist is no longer beginning only with an experience. They are beginning with an experience that has already been placed inside a psychological framework.

AI can accelerate this process because psychological terminology is readily available within the conversation.

A person describing difficulty concentrating might begin wondering about ADHD. Someone discussing a painful childhood memory may encounter language around trauma. Relationship conflict may quickly become a conversation about attachment, boundaries, manipulation, emotional availability, or communication styles.

None of these concepts are inherently problematic. Psychological language can help people articulate experiences they previously struggled to explain.

But labels also influence attention.

Once someone begins thinking, “This is what is happening to me,” they naturally start noticing information that supports that interpretation. Contradictory details may receive less attention. Ambiguous experiences begin fitting into an increasingly coherent story.

This is not an AI-specific flaw. It is a very human tendency. The difference is that AI can help build that story extraordinarily quickly.

 

AI Usually Meets the Story From One Side

Imagine telling a chatbot about an argument.

You remember what the other person said. You explain why it hurt. You describe previous situations that felt similar. You mention how frustrated you have become.

Everything the AI knows about the relationship comes through that description.

It wasn't sitting at the dinner table when the argument happened. It didn't hear the tone of voice. It doesn't independently know what happened earlier that day. It doesn't know which details were forgotten, softened, exaggerated, or considered too insignificant to mention.

Most importantly, it doesn't automatically know what the person telling the story cannot yet see about themselves.

This matters because therapy often becomes valuable precisely when the original explanation begins to feel incomplete.

A client may initially arrive believing their problem is that other people constantly disappoint them. Across months of therapy, they may begin recognising how difficult they find it to communicate what they need.

Someone may describe themselves as simply “bad at relationships” before noticing a recurring fear of vulnerability.

Another person may believe workplace stress is entirely caused by an unreasonable manager before gradually recognising their own difficulty setting limits.

Good therapy doesn't necessarily reject the client's original story. It creates enough space to examine it from different directions.

AI can help someone explore the story they provide. A therapist may eventually help them discover the story they didn't know they were telling.

 

Validation Can Help, but Therapy Cannot Always Agree

One reason people may increasingly take emotional experiences to AI first is simple: being responded to feels good.

When someone is upset, confused, angry, or ashamed, putting an experience into words can itself create relief. Receiving an organised response can make chaotic emotions feel more manageable. For someone who doesn't feel comfortable speaking to another person yet, that can be meaningful.

But there is an important difference between feeling understood and understanding what is happening.

Sometimes a person genuinely needs validation. Sometimes they need reassurance that an emotional response makes sense. At other times, growth requires something much less comfortable.

They may need a question they don't particularly want to answer. They may need to recognise their own role in a conflict. They may need to sit with uncertainty rather than immediately deciding who was right.

They may need someone to notice that the same explanation has appeared in five different relationships.

This is one reason therapy cannot simply become an experience designed around making someone feel better after every interaction.

A therapist's role isn't to win the client's approval by consistently confirming their interpretation. The therapeutic relationship needs enough trust for disagreement, curiosity, challenge, silence, and uncertainty to become possible.

Sometimes the most useful therapy session is the one that leaves someone thinking about a question for the next three days.

 

Therapists May Need to Ask About AI the Way They Ask About Google

There was a time when therapists didn't need to think much about what clients had searched online before appointments.

That changed when health information became widely available on the internet. Clients began arriving with symptoms they had researched, conditions they wondered whether they had, and terminology they had encountered online.

AI may create the next version of that shift.

In the future, asking how someone has been making sense of a problem may naturally include questions about AI.

Not because AI use itself is necessarily concerning, but because it may reveal something about the information environment shaping the client's understanding.

A therapist might discover that a client has spent weeks discussing their relationship with a chatbot. Another may have used AI extensively to understand panic attacks before seeking help. Someone else may be generating daily journaling prompts or asking AI to analyse conversations with family members.

Those interactions could become relevant therapeutic context.

The interesting question may no longer be simply, “Have you searched this online?”

It may become, “Who—or what—have you already discussed this with?”

For mental healthcare professionals, understanding these influences does not require becoming hostile toward AI. It requires recognising that clients' psychological narratives are increasingly being shaped outside the therapy room.

 

AI Could Actually Give Therapy a Better Starting Point

There is also a more optimistic possibility.

People often struggle to begin therapy because they don't know how to explain what they are experiencing.

They know something feels wrong, but putting it into words is difficult. They may feel embarrassed, confused, or overwhelmed by the idea of telling a stranger everything from the beginning.

AI may sometimes help people organise those thoughts before they seek professional support.

Someone might realise through an AI conversation that their sleep problems seem connected to workplace stress. Another person might recognise that they have been feeling persistently anxious rather than simply “busy.” Someone struggling with relationships may finally find language for a pattern they want to explore with a therapist.

In these situations, AI doesn't replace therapy. It helps someone arrive with better questions.

That could make the early stages of therapy richer. Instead of spending several sessions trying to identify what feels important, a client may arrive with reflections, examples, questions, and hypotheses they are ready to explore.

The therapist's role then becomes helping the client test those interpretations against the complexity of their actual life.

The goal isn't to erase what AI has already contributed. It is to move beyond it.

 

Therapy May Become the Place Where Digital Narratives Are Examined

For much of its history, therapy has helped people examine narratives.

"I'm not good enough."

"Everyone eventually leaves."

"I always have to be the responsible one."

"If I disappoint someone, they won't love me anymore."

These stories often develop through childhood experiences, relationships, families, culture, and repeated emotional patterns.

Now another influence is entering that process: algorithms.

The stories people tell about themselves may increasingly be shaped by search results, mental health content, online communities, social media terminology, and AI conversations.

That does not automatically make those stories false. But it makes examining them increasingly important.

One of therapy's future roles may therefore be helping clients separate useful psychological insight from interpretations they have adopted too quickly. A therapist may help someone ask whether a label genuinely explains their experience, whether another explanation deserves consideration, or whether an apparently neat narrative is hiding something more complicated.

In a world where technology can generate explanations almost instantly, therapy may become one of the few places where people are encouraged not to accept the first convincing explanation.

 

Technology Should Support the Relationship, Not Become the Relationship

There is a broader lesson here for the future of mental healthcare.

AI will almost certainly become more capable. Clients will use it. Therapists will encounter its influence. Mental health information will become easier to personalise and access.

Trying to build a wall between therapy and technology is unlikely to work.

A better question is where technology genuinely helps and where human judgement remains essential.

Technology can organise information, reduce administrative friction, support structured activities, maintain records, help with scheduling, and make continuity between sessions easier.

But the therapeutic relationship serves a different purpose.

Its value lies partly in the fact that another person is present—not simply another system producing an answer.

A therapist develops an understanding of someone over time. They can hold contradictions without immediately resolving them. They can recognise when the same pattern appears under different circumstances. They can challenge an interpretation while preserving the relationship in which that challenge becomes safe enough to consider.

The future of mental healthcare does not need to choose between intelligent technology and human therapy.

It needs to become much clearer about what each is actually good at.

 

The Third Person in the Room Isn't Going Away

A client entering therapy tomorrow may already have spoken to their partner, parents, friends, colleagues, Google, Reddit, Instagram, YouTube, and AI about what they are experiencing.

Therapists have never worked with clients in isolation from the world around them.

What is changing is the sophistication of the influences accompanying clients into therapy.

AI can now help someone interpret an experience before the emotional intensity has even passed. It can provide terminology, organise thoughts, suggest explanations, and reinforce or challenge particular perspectives. Those conversations may shape what clients notice, what they ignore, what they believe is wrong, and what they expect therapy to confirm.

That makes AI neither the enemy of therapy nor its replacement. It makes AI part of the client's context.

And perhaps that is how therapists will increasingly need to understand it.

The therapy room may still contain two people sitting across from one another. But the conversation will often carry the traces of many other conversations that happened before it.

The therapist's job will not be to pretend those influences don't exist.

It will be to help the client examine them.

Because when answers become easier than ever to generate, one of therapy's most valuable contributions may be helping someone decide which answers are actually worth believing.

---

FAQs

1. Can clients use AI before or between therapy sessions?
   Yes. AI can help clients organise thoughts, explore general mental health information, or prepare questions. However, it should not replace professional assessment, diagnosis, or therapy.

2. How can AI influence what clients bring into therapy?
   AI may introduce labels, explanations, or psychological concepts that shape how clients understand an experience before discussing it with their therapist.

3. Should therapists ask clients about their use of AI?
   It can be useful. Understanding whether AI, social media, or online information has influenced a client's thinking gives therapists more context about how their current narrative developed.

4. Can AI replace a mental healthcare professional?
   AI can provide information and support reflection, but it cannot replace the clinical judgement, therapeutic relationship, context, and professional responsibility provided by qualified mental healthcare professionals.

5. What if a client brings an AI-generated interpretation to therapy?
   It can become something to explore rather than automatically accept or reject. The therapist can help examine what fits, what may be missing, and whether other explanations deserve consideration.

6. Will AI change the future of therapy?
   Most likely. One major change may be that clients arrive with more information, stronger interpretations, and previous AI conversations, changing where the therapeutic conversation begins.