John Torous

The study found that Ash was better at sending users to crisis resources, like the 988 Suicide and Crisis Lifeline, and that its responses were less harmful than those from the GPT models.

John Torous, a psychiatrist at Boston's Beth Israel Deaconess Medical Center, sees a few problems with how that study was set up.

One of them, he says, is that the study only looked at cases that Ash flagged, so it's not clear if there were any risky conversations that it missed.

Slingshot says that the actual Ash product has other safety systems that were not part of the experiment.

It is not therapy.

But it's not just the mental health specific tools that have guardrails.

The general tools do too, and John Torous, the Beth Israel psychiatrist I referred to earlier, thinks the ones those models are putting in place are just as good.

There's some evidence that these bigger models have become so big, they've learnt so much, they've read so much, they can reason better, that they may be able to offer the same level, if not better level, of help.

The AI therapy industry is exploding.

John Torous, a psychiatrist at Boston's Beth Israel Deaconess Medical Center, estimates that there are at least 100 mental health-focused AI bots.

There's a huge range.

Some are coming from big companies specializing in mental health, like Headspace and Talkspace.

That's the early steps.

But we don't have that type of high-quality evidence to say a mental health AI is actually gonna be better.

John Torous, who I mentioned earlier, directs the Digital Psychiatry Division at Beth Israel Deaconess Medical Center.

He worked on a systematic review that's also not peer-reviewed, looking at how companies are evaluating their tools.

Many of these tools say user data is protected via encryption, but they also have optional settings where they save user conversations to train their models.

John Torous, a psychiatrist at Boston's Beth Israel Deaconess Medical Center, says this presents a risk that users might not fully realize.

You're probably not being given clear, direct informed consent where you are explained the risks and benefits and what happens to your data, and that is not gonna be a bright spot just for the field of how these companies built these models.

But for many of us, privacy is just one of those worries that tends to fall by the wayside.