Using An AI Persona To Impersonate Your Therapist So You Can Get Inside Their Head

AI And Mental Health

As a quick background, I’ve been extensively covering and analyzing a myriad of facets regarding the advent of modern-era AI that produces mental health advice and performs AI-driven therapy. This rising use of AI has principally been spurred by the evolving advances and widespread adoption of generative AI. For an extensive listing of my well over one hundred analyses and postings, see the link here and the link here.

There is little doubt that this is a rapidly developing field and that there are tremendous upsides to be had, but at the same time, regrettably, hidden risks and outright gotchas come into these endeavors, too. I frequently speak up about these pressing matters, including in an appearance on an episode of CBS’s 60 Minutes, see the link here.

Background On AI For Mental Health

I’d like to set the stage on how generative AI and large language models (LLMs) are typically used in an ad hoc way for mental health guidance. Millions upon millions of people are using generative AI as their ongoing advisor on mental health considerations (note that ChatGPT alone has over 1 billion weekly active users, a notable proportion of which dip into mental health aspects, see my analysis at the link here). The top-ranked use of contemporary generative AI and LLMs is to consult with the AI on mental health facets; see my coverage at the link here.

This popular usage makes abundant sense. You can access most of the major generative AI systems for nearly free or at a super low cost, doing so anywhere and at any time. Thus, if you have any mental health qualms that you want to chat about, all you need to do is log in to AI and proceed forthwith on a 24/7 basis.

There are significant worries that AI can readily go off the rails or otherwise dispense unsuitable or even egregiously inappropriate mental health advice. Banner headlines in August of this year accompanied the lawsuit filed against OpenAI for their lack of AI safeguards when it came to providing cognitive advisement.

Despite claims by AI makers that they are gradually instituting AI safeguards, there are still a lot of downside risks of the AI doing untoward acts, such as insidiously helping users in co-creating delusions that can lead to self-harm. For my follow-on analysis of details about the OpenAI lawsuit and how AI can foster delusional thinking in humans, see my analysis at the link here. As noted, I have been earnestly predicting that eventually all of the major AI makers will be taken to the woodshed for their paucity of robust AI safeguards.

Today’s generic LLMs, such as ChatGPT, Claude, Gemini, Grok, and others, are not at all akin to the robust capabilities of human therapists. Meanwhile, specialized LLMs are being built to presumably attain similar qualities, but they are still primarily in the development and testing stages. See my coverage at the link here.

The Use Of AI Personas In Mental Health

In the AI and mental health realm, there are lots of ways that AI is being used, including by invoking AI personas. An AI persona consists of telling an LLM to pretend to act in some particular way. A budding therapist might tell AI to pretend to be a certain type of generic client who has a specific mental health disorder. The newbie therapist can practice their skills on the AI persona. It’s a no-harm, no-foul situation. If the therapist messes up, the AI persona isn’t a real human being. Meanwhile, the therapist improves their craft and learns valuable lessons.

In the case of psychological research, researchers might invoke dual AI personas: one as a therapist and the other as a client. The AI personas interact as though they are having an everyday therapeutic discussion. Afterward, the researchers can analyze how things went and explore various therapeutic methodologies. The AI can also be improved in mental health capabilities by telling the AI to learn lessons from the interactions that took place.

For more on this topic, see my discussion at the link here.

I have previously described in detail how to appropriately invoke an AI therapist; see the link here. You give a prompt to AI that identifies the nature of the therapist, such as what their mental health specialty consists of, their experience level, and so on. Creating AI personas as clients is also something that should be done systematically. You can invoke an AI persona client that is young or old, has a specific mental disorder or multiple disorders, etc. See my explanation of how to suitably invoke AI persona clients at the link here.

Clients Invoking AI Persona Therapists

There is a wild twist associated with the use of AI personas in a mental health context. Clients who regularly see a therapist are at times opting to craft an AI persona that is likened to their human therapist. This might seem an odd practice and perhaps even disconcerting. Shouldn’t the person just see their therapist and not try to create an AI version? What could be a reasonable basis to make an AI persona of your therapist?

Let’s closely examine the matter.

First, a person might establish an AI persona of their therapist so they can prepare to see the therapist and be suitably ready for scheduled therapeutic sessions. This person believes that they can maximize the precious and limited time with their human therapist by getting ready beforehand. No sense in wasting billable time. It would seem logical to consider what topics to cover and what you are hoping the therapist will advise you on. If you only have a 1-hour session per week, you’ve got to squeeze the juice out of each minute.

Second, after meeting with your human therapist, it might be useful to do a debriefing by dipping into an AI persona of the therapist. A client oftentimes is puzzled why their therapist brought up certain topics. The client might wonder what the therapist was driving toward. Sure, it would be best to simply ask the therapist, but access to the therapist post-session might be limited or expensive. Using an AI persona would be a seemingly sensible second-best option.

Third, a client might be seeking advice throughout the week while outside of a session, but have no immediate means of communicating with their therapist. They must bottle up their questions or issues. They might be unsure of making an important decision and feel that without consulting the therapist, they cannot do so. But waiting until a once-per-week session could be difficult or nearly impossible. Instead, they log into AI and ask the AI persona that mimics their therapist.

Those and other perfectly sensible reasons are aboveboard. I mention this as being aboveboard because the first assumption is that if someone crafts an AI persona of their therapist, they must have some devious motives in mind. It sounds creepy. You can see that there are reasonable purposes that justify this.

The Downsides Are Notable

As with most things in life, there is never a free lunch. The use of AI personas as a pretense of a human therapist is fraught with many downsides. Though the above upbeat reasons are valid, you must compare them to the weaknesses and disconcerting aspects.

One of the main downsides is that an AI persona is not going to be a realistic representation of a human therapist. No matter how hard you try to set up an AI persona, rest assured, it won’t be on par with the modeled human being. The AI persona will say things that your therapist might never have said. The AI persona could end up giving you guidance that is completely the opposite of what the human therapist would tell you.

Another related concern is that the client might anchor on the AI persona. A kind of tug of war ensues. The client goes into a session and wants to debate the human therapist by saying that the AI persona told them something different from what the therapist is saying. When the therapist suggests that the client should disregard the AI persona, the client becomes confused and upset, arguing that the AI persona is (supposedly) the same as the therapist.

You see, it makes no sense to the client that the human therapist and the AI persona are not fully aligned. Furthermore, a client can get mentally muddled with what the human therapist advises versus what the AI persona advises. Imagine things this way. The AI persona tells the client to do Z. The human therapist tells the client to do R. In the mind of the client, they believe that the human therapist told them to do Z. It becomes an interchangeability issue that the client cannot keep straight.

How Such AI Personas Get Made

You might be wondering how someone would make an AI persona of their therapist. It can be extraordinarily easy. The approach can range from simple to complex. It all depends on how far someone wants to take this.

The first thing to do would be to feed into the LLM whatever might be readily available regarding the therapist. It might be the bio of the therapist. It could be content from their website that describes their background and philosophy about therapy. Perhaps the therapist has a blog or has posted their methodological preferences online. All of this is fodder for crafting the AI persona.

Doing so requires using suitable prompts and leaning into retrieval augmented generation (RAG) in-context-learning techniques, see my explanation at the link here. No special AI developer skills are required. This is done by using user-level capabilities. No special training is needed.

You can go deeper. A client might opt to enter their personal notes from their therapy sessions. If they somehow have transcripts, they might feed those into the AI. The aim is to go beyond the usual outside information about the therapist and seek to use the insider insights that the client has gleaned.

Despite all of this and no matter how deep you think you’ve gone, it is crucial to realize that the AI persona will not be a proper representation of the therapist. You are still working with computational and mathematical simulation. Nothing more.

Therapists Protecting Themselves

I’ve predicted that we are heading to a legal quagmire of people setting up AI personas of human therapists and then sharing those AI personas with other people. In a sense, it is a rip-off of the human therapist.

If anything, therapists are bound to create an AI persona of themselves, doing so as a means of expanding their therapeutic practice (see my discussion at the link here). They do so using their own IP (intellectual property). Having a client or someone else opt to do the same, and then potentially profit or otherwise exploit that, has significant legal and ethical ramifications.

More Vital Factors

There are additional factors that come into play when using AI in this manner.

Suppose the AI provides an oddball answer that seems convincing and advises the person to do something bizarre. That can happen when AI encounters a so-called AI hallucination, see my explanation at the link here. The AI can have a kind of confabulation, whereby it produces a plausible-looking answer that is factually incorrect. It looks right, but is misleading or inappropriate.

Privacy issues also enter into the picture. Most people assume that their use of AI is considered private and confidential. Nope, that’s rarely the case. The AI makers typically stipulate in their online licensing agreements that any chat you have with the AI can be inspected by their developers. Furthermore, your chat can be used to further train the AI. Bottom line is that your privacy and confidentiality are not guaranteed, and potentially, you are opening yourself to privacy intrusions.

The World We Are In

Let’s end with a big picture viewpoint. It is incontrovertible that we are now amid a grandiose worldwide experiment when it comes to societal mental health. The experiment is that AI is being made available nationally and globally, which is either overtly or insidiously acting to provide mental health guidance of one kind or another. Doing so either at no cost or at a minimal cost. It is available anywhere and at any time, 24/7. We are all the guinea pigs in this wanton experiment.

The reason this is especially tough to consider is that AI has a dual-use effect. Just as AI can be detrimental to mental health, it can also be a huge bolstering force for mental health. A delicate tradeoff must be mindfully managed. Prevent or mitigate the downsides and meanwhile make the upsides as widely and readily available as possible.

A final thought for now. There is a bit of a joke going around that pertains to making an AI persona of your therapist. The joke goes this way. After switching to using the AI persona rather than the human therapist, the person decides to make an AI persona of themselves. Consequently, they merely have their AI persona of themselves that confers with the AI persona of their therapist. They no longer need to engage in human-to-human therapy (or so they believe). Just let the AI handle it all. Perhaps it’s not quite as laughable as it might seem.