Could AI Avatars Be Used in Clinical Psychology Training? A Psychotherapy Podcast Episode.
With the constant march forward of artificial intelligence and its continuing impact on our lives, we have to question how it will impact clinical psychology. In this psychology podcast episode, I provide a potential answer because I reflect on an experiment that I took part in during my Masters of Science. If you’re interested in cyberpsychology, the future of mental health training and careers in psychology, then this will be a great episode for you.
Today’s psychology podcast episode has been sponsored by Beyond The Lecture Volume 8: Thoughts on Clinical Psychology, Mental Health and Psychotherapy. Available from all major eBook retailers and you can order the paperback and hardback copies from Amazon, your local bookstore and local library, if you request it.
Could AI Avatars Be Used in Clinical Psychology?
I mean this as a joke but one of the issues with being very popular amongst the lecturers at my university is that I can be walking round the psychology building whilst working as a Student Ambassador, then I can run into a lecturer and suddenly be recruited for a fun psychology study.
Normally, Masters students never have to partake in research studies but this one sounded fun, the Masters students needed clinical people for their sample and I am always happy to help others.
In Clinical Psychology Reflections Volume 7, I wrote about my experience with the first part of this study looking at how an artificial intelligence chatbot was pretending to be a client with depression and anxiety. Then I had to interact with the chatbot in a mock psychological assessment situation. In that reflection, I wrote about the strengths and weaknesses of the chatbot and whether I believed it should be used or not in future clinical psychology training.
Today I had the second part of the study that used a similar structure in terms of I had to conduct a mock psychological assessment with a man with depression and anxiety. Yet this time, the artificial intelligence was a scarily realistic avatar that looked exactly like a real person.
For me, it is just scary how realistic AI is these days and how you can create a real-looking person that I wouldn’t be able to tell was artificial intelligence with the press of only a few buttons.
For the next ten minutes, I needed to follow the CBT cross-sectional formulation design to help me gather information for my psychological assessment. I asked him about his home situation, his thoughts, feelings, bodily sensations and suicide, alcohol and substance abuse risk. As well as whether there were any situations that triggered his low mood or anxiety the most.
Another piece of context for the client is that he was an architecture student from an unknown country, and part of his mental health difficulties is that he is struggling to adapt to student life in the UK because he feels lonely.
One of the first issues with the AI avatar is that when I asked him about his home culture he couldn’t answer. He only told me about “back home” and “when I moved here”. I asked in different ways because I understand in the real world, a client might feel ashamed or concerned about revealing their home country.
The AI avatar still didn’t answer so I feel like this was a critical part of background information that I was missing. As well as depending on where the client was from, their home culture probably would have impacted my questions and my ideas for future psychological treatment.
After the ten minutes were up, the researcher came back into the room and she asked me a bunch of open-ended questions about the ethics, advantages and disadvantages of the AI avatar, how I thought it could be improved and so on.
What Are The Positives Of The AI Avatar?
It turns out there are advantages to having an AI Avatar help you in clinical training. I found it helpful that the AI Avatar looked scarily realistic and his responses were very similar and genuinely how I believe a person with depression would respond to my questions. I’ve given very similar answers in the past, especially during my counselling assessment for my specialist rape counselling when I was experiencing mild depression.
That turned into severe depression a month later.
Therefore, it’s useful to have a realistic AI Avatar for trainee clinical psychologists because it allows them to practice clinical skills in a situation similar to the real world. A situation where they would need to carefully word questions, react to what the client was saying and monitor body language.
I’ll talk more about body language in the disadvantages section.
In addition, having an AI Avatar could be very useful in clinical training because it means you have to think on your feet. Since the AI Avatar is right there in front of you, you need to think about your responses, practice your active listening skills by showing that you understand what they said and you need to follow up on from their response.
Yet unlike the AI chatbot, you actually need to do this quickly like you would in the real world otherwise you get into awkward silences. Even though in the real world, silence isn’t always a bad thing.
On the whole, an AI Avatar might be useful for helping clinical trainees to develop these interpersonal and response skills.
Building upon this advantage, a benefit of the AI Avatar over the chatbot is because I was typing with the chatbot, after a while, I wasn’t thoughtful about what I asked and I just asked it. I didn’t care about my wording or my phrasing because typing just isn’t realistic, it’s boring and it was taking a long time to type a response to what the client had just said and my next question.
Nonetheless, when it came to the AI Avatar, I was finding myself being a lot more thoughtful about what I asked because the “client” was right in front of me, blinking and smiling at me.
I had to act as if this was a very real client and I was speaking so it was easier to practise those active listening skills and phrasing my questions in a kind and compassionate way.
Granted, I will be the first to admit that I need to improve this skill. I am not great at knowing how to phrase my questions in certain ways.
Leading me onto my last point, by using the AI Avatar, I was able to identify some areas for improvement. For example, I need to practice wording my questions and responses in a careful and considered manner so I don’t offend future clients and risk damaging the therapeutic alliance.
On the whole, I think there are several advantages of an AI Avatar being used in clinical training. For example, AI Avatars can be useful because they look and give realistic responses that can help trainee psychologists get in more practice outside of their placements. This will allow them to improve their clinical skills even more.
In addition, AI Avatars can help trainee psychologists to identify areas for improvement, think more carefully about their responses amongst other benefits.
What Are The Negatives of The AI Avatar?
I’ve already mentioned how I didn’t like that the “client” didn’t know where they were from and some of the background information was missing. Yet there were other issues too. For example, I felt that a major issue of the AI avatar was that he looked real, his words were similar to those of a depressed person, but his body language was completely wrong for a depressed or anxious person.
Of course, you could argue that the client with depression was feeling good that day, that he was putting on a brave face and he was trying to trick me as a clinical psychologist that he was doing better than he was. This is possible in the real-world, especially if a client doesn’t want to be there and doesn’t want psychological support.
This AI Avatar wasn’t doing that.
If he was trying to fake his reactions then I ask you the following questions:
· Why was he giving me responses about his presenting problems and other factors that reflect depression in the real-world?
· Why was he encouraging us to solve certain difficulties in therapy sessions?
· Why didn’t he mention he was here at someone else’s request?
Therefore, I really didn’t like how the AI avatar didn’t act or show body language similar to that of someone with depression. They didn’t look sad, they didn’t look nervous and they didn’t show that every movement or action was taking a lot of effort.
If the AI avatar had severe depression like it was trying to convince me then he would have showed these behaviours.
Moreover, this is important to note because in a real-life psychological assessment, a psychologist would be reading and making notes of the client’s body language. At the moment, I’m doing an essay assignment where I need to create a psychological treatment programme based on a letter with the results of this client’s psychological assessment. The very first section of this letter is about the client’s body language during the assessment.
If the AI Avatar does not use body language, how am I meant to improve my ability to read and understand client body language?
I can’t.
In addition, when I compare the AI Avatar to the AI chatbot I mentioned in the first part of the study, at least the chatbot mentioned what the client was doing when I asked questions. The chatbot mentioned how the client was picking at a loose piece of their hoodie sleeve, they were biting their lower lip and they were showing other signs of body language.
Surely it is easier for an AI Avatar to show body language, so why didn’t it? (besides it not being programmed too).
Another minor issue along the same lines is the tone of voice was slightly wrong, and the small smile the AI Avatar always had was very off-putting. It seemed wrong that the AI Avatar was always smiling even when he was telling me about his depression, how low he felt and how he had considered suicide a few times.
It wasn’t realistic in that regard and it was really, really off-putting.
Therefore, I would highly recommend that if AI Avatars are going to be used in future clinical psychology training then they need to, at the very least, sound like they have depression, sound like they have low mood and they need to not smile as much.
When I was severely depressed in the latter half of 2024, I was not smiling too much, you could hear the depression in my voice and I was not always even remotely happy.
These AI Avatars are, and it just doesn’t match up with what the AI Avatar is saying.
Finally, besides from being a little repetitive because sometimes regardless of the question I asked, he gave me very similar answers to previous questions so I wasn’t always able to get the information I needed for my psychology assessment. One issue I didn’t have with the AI Avatar was that he was meant to be experiencing depression and anxiety, he got the depression, he didn’t get the anxiety.
He didn’t tell me about having sweaty hands, feeling panicked and overwhelmed and he didn’t tell me about any situations that he felt anxious in.
As a result, this leads me to one of the disadvantages I mentioned to the researcher, I do not think that AI Avatars will be useful in training clinical psychologists to support those with more “complex” mental health conditions. For example, psychosis, those with personality disorders, trauma responses and those with hallucinations and delusions.
If an AI Avatar that is meant to show depression and anxiety can only show one condition that we know a lot about, then I am concerned how an AI Avatar will manage to accurately represent a more complex condition. Since if we take obsessive-compulsive disorder for example, it is not only handwashing and being clean. OCD can encompass a lot more.
Here is a small extract from an episode of The Psychology World Podcast focusing on OCD:
“DSM-5 includes other information in the diagnostic criteria. For example, the obsessions or compulsions take up more than one hour a day, cause impairment for the person and/ or they cause clinically significant levels of distress. Then there is the typical DSM-5 caveat that Obsessive Compulsive Disorder needs to be the best explanation for what the client is experiencing and their symptoms cannot be better explained by another mental health or medical condition or a substance.
When it comes to the obsessions themselves, the specific details of them can vary wildly between different people with Obsessive Compulsive Disorder. They can include thoughts round contamination, a desire for order or they can be taboo thoughts around harm to themselves or others, sex and/ or religion.”
As you can see, there is a lot of variation in OCD so I think it would be very difficult to accurately programme an AI Avatar to show this range of behaviours. Then I think it is “dangerous” or “unhelpful” to our future clients if we train trainee psychologists with an Avatar that reinforces and only shows stereotypes.
Just a thought.
Would I Recommend The AI Avatar For Clinical Training?
Personally, with the rise of artificial intelligence and how it is impacting more aspects of our lives. I strongly believe that at some point artificial intelligence will be a part of clinical psychology whether we like it or not. I already talked about artificial intelligence and chatbots to some extent in clinical psychology because of mental health apps in my book Could Apps Improve Our Mental Health?
However, when it comes to whether I would recommend AI Avatars in clinical training. I would say not at this current moment in time but in the future, I think there is definitely a place for AI Avatars.
I can see AI Avatars being very useful in the beginning of clinical training because it will help trainees develop and extra practise in those core competencies that they will be using every single day with clients. They’ll be able to practise their active listening, communication and relationship-building skills with the AI Avatars.
Also, AI Avatars will be useful because instead of trainees only being able to practise these skills with real-world clients during business hours. Trainees will be able to practise whenever they want.
However, these AI Avatars are not perfect and they have limited uses at the moment. Before I highly recommend these avatars for clinical training, they need to be more realistic in the terms of body language and voice tone so trainees can develop their body language monitoring skills too.
In addition, the AI Avatars need more background information, especially if they are meant to be someone who has recently moved to the region or country because their culture of origin could be critical to understanding the social factors underpinning their mental health condition.
Then there are other disadvantages too that I mentioned earlier in the chapter.
I think there is a place for AI Avatars and there is great potential for this technology in training so our trainees can become even better at helping our amazing clients. I just don’t think we’re there yet.
What do you think?
I hope you enjoyed today’s cognitive psychology podcast episode.
If you want to learn more, please check out:
Beyond The Lecture Volume 8: Thoughts on Clinical Psychology, Mental Health and Psychotherapy. Available from all major eBook retailers and you can order the paperback and hardback copies from Amazon, your local bookstore and local library, if you request it.
Have a great day.
Clinical Psychology Reference and Further Reading
Whiteley, C. (2026) Beyond The Lecture Volume 8: Thoughts on Clinical Psychology, Mental Health and Psychotherapy. England. CGD Publishing.
I truly hope that you’ve enjoyed this blog post and if you feel like supporting the blog on an ongoing basis and get lots of rewards, then please head to my Patreon page.
However, if want to show one-time support and appreciation, the place to do that is PayPal. If you do that, please include your email address in the notes section, so I can say thank you.
Which I am going to say right now. Thank you!
Click https://www.buymeacoffee.com/connorwhiteley for a one-time bit of support.
Comments