Artificial Intelligence Enters Therapy
Emerging AI medicine is about more than just robot surgeons — it's redefining healthcare as we know it, and making therapy more effective than ever.
When you think of AI, you might imagine some futuristic scenario like in I, Robot or Westworld, where intelligent machines walk among us and perhaps try to kill us all.
In reality though, artificial intelligence has been steadily creeping into your life for many years now. The algorithms that sort your social media feeds, autocomplete sentences for you, or de-blur your the photos on your phone, are common examples of AI that we’re already taking for granted.
AI chess players have been beating grandmasters for so long that the current models are now learning to make more human mistakes, in order to understand us better.
But of course, this is just what we encounter each day. New uses for AI are appearing virtually every day, for everything from art to medical care.
Take for instance, this wild post from the MIT Technology Review. Companies are now starting to use Artificial Intelligence to analyze the transcripts of therapy sessions, leading to better outcomes for patients.
We are entering a new era in mental health care, one in which AI plays a major role in diagnosis and treatment. This is not to say that AI will replace human therapists, but AI can complement and augment the work of therapists in a number of ways.
Using AI for Personalized, Data-Backed Guidance
Certain types of conversations in therapy sessions lead to better outcomes, and tech companies are now using AI to consult therapists, in order to improve their quality of care.
In one finding, it looks like the amount of a time in a session that a therapist engages in idle “chitchat” or “rapport building” is negatively associated with improvements in a patient’s mental health — in other words, less chitchat = more therapy = better results. Go figure.
The average success rate of therapy for depression is 50% — a coin flip — but one AI therapy consultant is reporting improvements with AI have it up to 62%.
Similarly, success for anxiety treatment jumped from 58% to 73%.
Bear in mind, even a 1% improvement for a major care provider means tens of thousands of people are now free from mental illness who otherwise would not have been.
The algorithms also found that patients expressing a desire to change, or reflecting on ways to change, was strongly associated with positive outcomes. Now engineers are trying to improve the algorithm to detect what therapists might be saying or doing to elicit these kinds of responses
On one hand, this might raise potential privacy risks if not properly handled. But then again, being as we are in the midst of a mental health epidemic, maybe it’s about time we applied some hard data to improving our quality of mental health care.
The Future of AI in Medical Care
As anyone who’s heard the AI paperclip maker thought experiment knows, if artificial intelligence wipes out humanity, it probably won’t be from some malicious bot hellbent on ending the human race. More likely, the AI will do what it was programmed to do, follow the instructions we give it, even if these instructions eventually require it to do something that kills us off as a consequence.
On the other hand, sometimes the risks of AI are not so much about the AI itself, but how humans might use it. Take for instance, AI algorithms that can be used to quickly determine someone’s race using X-rays, since some races are more vulnerable to certain diseases than others). This has some helpful medical applications, but could also be abused, such as being used by insurance companies or advertisers to discriminate based on race.
AI is already making incredible contributions to health care however, such as predicting breast cancer, identifying rare diseases in children, or — more impressively in some ways —optimizing the logistics of medical treatment (such as supply chains). This last project is already saving lives and reducing medical costs by offering counterintuitive solutions to modern healthcare problems around the world.
So what happens if we can harness AI as a force for good, giving humans the ability to solve our most difficult problems? It looks like we might find out in our lifetime.
Speaking of therapy, if you liked today’s post, you might enjoy this one too: