Naomi Alderman

went out for a walk, open-mouthed in the sun. I definitely swallowed a fly.

(I do know why. Because it flew right into my throat and then what are you going to do? Tried to spit it out, no good.)

some of the most useless, and in fact harmful therapy I ever had was with someone who was SIMPLY TOO STUPID to understand what I was saying to her. a real slow-blinker.

(there was other harmful therapy with someone not stupid at all which is a real story, boy howdy. and some very good therapy.)

it's an actual shanda.

that stuff *can* be useful, in my experience, about 10-15 years into real therapy where you've gone "oh YEAH, probably my childhood DID have an effect on me, and THAT'S why I'm like this". and done crying and being angry. then one day you go "hmm, maybe I could shift a bit"

But I don't know. That is just reckons. It would need to be actually studied.

I suspect that LLMs might turn out to be slightly *less* harmful than an idiot reading a checklist at you.

Because at least you didn't open your heart to another human who completely failed to give a shit or even basically be able to understand what you were saying.

I would very much like to know how many people in crisis were seen in this kind of situation and ended up harming themselves, and comparing that to the statistics about LLMs and harm.

That is not "LLMs and harm is fine!" It is that I am wildly enraged about how bad manualised CBT is.

I think a great deal of harm is done by the NHS pretending that CBT delivered in this way is the gold standard. We just need to say out loud and frequently: the NHS does not cover mental health treatment except for those who are in severe crisis.

Photo of Silvius the little Latin mouse dressed as The Very Hungry Caterpillar for World Book Day

#fabulamurina (mouse story) 491
hodie diēs librorum orbis terrarum est (Today is World Book Day). Silvius simulat personam e libro esse - eruca ieiunissima est! (Silvius is pretending to be a character from a book - he is The Very Hungry Caterpillar!)
#WorldBookDay

To say out loud: there is great treatment available but unfortunately it is only provided now in the private sector. We are sorry. And even better if we could also say: the NHS has a service to help you identify low-cost therapists and to work out which therapist is right for you.

I think a great deal of harm is done by the NHS pretending that CBT delivered in this way is the gold standard. We just need to say out loud and frequently: the NHS does not cover mental health treatment except for those who are in severe crisis.

yes my strong feeling for a long time is that the NHS just needs to be able to be honest about this. we don't have the resources to give competent free-at-the-point-of-use psychotherapy to everyone who needs it in the UK. we barely have the resources to cover those in extreme crisis.

I am very much here as a safe space for everyone’s incredibly bad experiences with idiots applying CBT worksheets at them.

I think every clever/sensitive/both person I know in the UK has had some experience like this. It is proper medical harm.

yes this too. we all yearn to be held in a perfect memory, don't we? (maybe just me, or maybe a widespread but not universal wish.)

this is an attraction of God: that all the facts of your life are known and remembered and you are still loved.

yeah. therapy with a human person involved is not always a 'gold standard' especially if that human person is an idiot.

I know almost nothing about Korean society tbh and have never been. I wonder why it works so well there. Whether there are effects of being right next to North Korea that lead to a general "we are constantly on a war footing & the population must be healthy" social cohesion?

yeah, there were times in my life when I was literally deciding between "can I afford to get the bus somewhere for an evening out, or do I need that money for therapy?" and ending up going "no I am saving it for therapy".

I would like to do an in-conversation with an experienced therapist where we talk through a harm-reduction approach to using AI for therapy.

I feel like - as with writing - there's a lot of collegiate pressure not to give an inch or even acknowledge that people are using it like this.

well I am desperate to understand how this works! I have never received any even vaguely competent therapy from a health service, always paid for it myself both in the UK and the US

I am SURE this is true.

Computerised CBT is dreadful. I know so many people who have experienced significant harm from inappropriately prescribed, badly-applied and badly-delivered CBT.

It is *so easy* for it to turn into "this is all your own fault/you are irretrievably broken"

I would like to know how this works! I certainly know PRECISELY ZERO about the Korean health system

(and fwiw I'm in the UK and we have the NHS and it's great in many ways - and has its problems - and 'providing mental health help as needed' is certainly not a solved problem here. I'm not aware of anywhere that it's a solved problem tbh.)

I don't disagree that this is what we should do. but we live in an imperfect world and I'm considering what *I* could possibly actually do that would help.

I think these are just two different questions. ideals and pragmatism both have their place. I'm in a pragmatic place in this right now.

I know so many people, clever successful people, who thought for a decade or more that therapy couldn’t help them because they had six shit sessions with someone reading off a worksheet at them. And what they needed was 10 years committed work with a brilliant well trained person.

Like, I don’t think a brilliant therapist probably should be paid £20 an hour. It’s a real problem for the NHS, and I don’t think that the way it handles things right now – which is basically pretending that a few sessions of CBT is enough – is helpful for people either.

I think it’s also just incredibly expensive. I saw someone wonderful for 25 years at a cost of on average about £150 a session. That changed my life, and she bloody earned her money. But it is a huge amount of resource, on a population level.

And I think it would probably also be very very good to investigate best practices and to work long-term on creating an LLM that is specifically designed, and has all the many safeguards needed, for this type of work. It’s not just a casual use, it’s the most common use.

It would be great to do some formal studies in best methodology rather than just going on vibes. It would be great for some large therapeutic association to do a study in a combined therapy: using LLMs to augment work done in the therapy room.

I think I feel certain that at least: as so many people are using LLMs for therapy, it is incredibly important to have loud public conversations about harm reduction. How to use it in ways that are least harmful and most likely to be helpful.

yes, and I think the evidence actually shows that the quality of the human relationship with the Therapist matters much more than any specific treatment modality or school of theory. If it works, it’s a form of reparenting.

Yes, like everyone who’s taken therapy seriously I’m sure, some of the most productive sessions I ever had were ones where my therapist pushed back against what I was saying. I actually have my LLMs set up to always give a steel man counterview, basically for this reason.