Is something better than nothing?
There’s a lot to put us off talking about AI therapy. Like the way five minutes into the what’s-your-problem-with-chatbot-therapy-debate you could find yourself wanting to trade in other people for a conversational robot puppy. Walking the AI literacy tightrope involves the unsettling movement between enjoying the professional bun fight over algorithmic bias on Linkedin while at the same time experiencing PTSD every time you have to explain to your mum what an algorithm is.
Last week I had the strangest experience of genuine relief watching a US Congress committee looking at AI Chatbots and mental health and the thoughtfulness with which they handled it. This conversation was tucked away under the potentially unlikely structures of the US House Committee on Energy and Commerce, which when you think of it does make sense as the UberTherapy business model shifts to a sector of e-commerce. In the proceedings we hear from some professionals at the top of their game, summarizing key institutional data and the evidence base from prestigious Ivy League pilots around providing digital and AI technologies into the field of psychiatry and mental health management systems.
Alienating as a governmental committee can be, it was peppered with interventions from Dr Maralyn Wei who calmly provided balm to its political nervous system by starting with the basics. She lays out three types of chatbots operating in the field of mental health - general purpose chatbots, AI companions, AI therapy chatbots specifically designed for mental health each with its own regulation and oversight, most of it vague. She reminds us that chatbots are designed to provide validation, direct advice and to avoid complexity because it is on this basis that AI therapy chatbots can claim 50-75% recovery-as-improvement data. This data then goes on to build the evidence base for why chatbots have culturally secured a place as legitimate agents in the therapy game. She also lays out cooly the risks associated with chatbot use centering around emotional and relational risks, reality testing, crisis management and systemic risks that consider the contexts within which people are seeking help from AI.
Dr Wei then simply says that the risks are considerably higher to people who are lonely, isolated or experiencing complex and systemic problems in their lives. And with that simple statement you see the problem.
The research race underway to get the data on AI chatbots already indicates a range of factors, not easily reduced to generalisations about generational splits or digital efficacy, rather drawing us back towards the wicked problems of mental health, that the people most likely to need real therapy are the least likely to be able to access it. Despite this, almost all of the research around AI Chatbots reflects a familiar story about what happens when we hit a crisis without the possibility of accessing the people and resources within our healthcare systems required to manage it. And there’s the rub. The people most likely to turn to this form of support are the most likely to be harmed by it.
It is true to say that not everyone is designed to stare into the abyss, but one of the problems with the something-is-better-than-nothing argument supporting ChatGPT Therapy-Lite is that it plays upon our ambivalence of having relationships with other people. Life is often about what the alternatives are, and if the offer is a burned out NHS therapist after a 12- month waiting list for guided self- help and endless mood questionnaires, it is legitimate to ask if a digital something is better than the emerging clinical nothing. But what AI systems cannot do, by design, is to contextualize, to think critically and to resolve conflicts; a profound political deficit in the emerging interlock between authoritarian systems and AI technologies, and an unsafe environment for therapeutic work. Not saying its impossible to get help from a Chatbot, just that its’ convenience comes at a potentially hidden and high cost.
Throughout the writing of this UberTherapy I have read and re-read Robert Money- Kyrle’s 1960s’ papers about a psychoanalytic framing of the political self, partly because of the nuance in among the big strokes of his writing, and partly because it was one of the first writings that made it possible for me to respect my political mind while having analysis. There are three different but interlinked facts of life presented in Man’s Picture of His World and throughout UberTherapy I try to formulate them in relation to the AI world that we are now living in. The proposal is that digital therapy and the use of an AI therapist can be seen as a defence against the three intertwined facts of life, unavoidably writ large in therapeutic processes and for anyone who needs to do the internal work of repair. First that we are dependent on others to grow – not to be ‘connected’ or to swipe left, but to allow ourselves to need others and to navigate that. Second that we are not self- replicating and come into being through a creative act that we are excluded from, and this requires that our aspirations to be self-sufficient are kept in their place. Then we die – the third fact is that time is not infinite and all things end beyond our control, despite the seduction of being able to relate on-demand and ghost people before they leave us. And it is this industrial context of AI therapy that means that when we nudge each other into a ‘could-be-worse’ defence of using chatbots that we become complicit in reaching the logical conclusion of platformization, which is therapy becomes a sub-division of e-commerce.
UberTherapy has happened by us not asking the right questions in our standardized self-reported questionnaires and not challenging the industrial and financial architecture behind the new business of mental health. It is in this way that we stand at the tipping point in the ‘race to the bottom’ that UberTherapy represents. And it can only be by protecting our ability to really say what’s on our minds that we can work through whether, for each of us, something is actually better than nothing.
You can buy a copy of UberTherapy: The new business of mental health by BUP here
@survivingwork.bsky.social @survivingwk
@UberTherapy.bsky.social @ubertherapies