Misty Williams checks into the emergency room at Cedars-Sinai Medical Middle once in a while for therapy of debilitating ache from sickle cell illness, which causes crimson blood cells to stiffen and block the circulate of blood.
After ache medicine and hydration are ordered, the 41-year-old Los Angeles resident makes an uncommon request: entry to a digital actuality headset with a synthetic intelligence-powered chatbot that may stick with it a dialogue together with her.
With the headset on, Williams finds herself in a digital backyard, butterflies drifting round her. A humanoid robotic greets her with a soothing feminine voice.
Affected person Misty Williams, carrying an Apple Imaginative and prescient VR headset, has a session with an AI-powered “robot” named Xaia at Cedars-Sinai.
(Genaro Molina / Los Angeles Instances)
“Hi, and welcome. My name is Xaia, and I’m your mental health ally,” it says. “How can I help?”
After a session, Williams’ ache eases and her thoughts is calmer.
“Mentally and physically, I feel more at peace,” Williams mentioned.
Xaia (pronounced ZAI-uh) is only one of many ways in which synthetic intelligence expertise is barreling its method into the burgeoning sector referred to as digital well being.
Digital well being startups utilizing AI accounted for an estimated $3.9 billion in funding in 2024, or 38% of the sector’s complete, in accordance with the digital well being advisory agency Rock Well being. Psychological well being was the top-funded medical space, drawing $1.4 billion.
Main medical establishments in Los Angeles are embracing the pattern. UCLA Well being is utilizing AI to assist medical doctors catch strokes quicker, scale back hospital re-admissions and spend extra time with sufferers by automating medical notes, mentioned Paul Lukac, chief AI officer.
Keck Drugs of USC plans to supply staff a industrial AI chat software to help stress administration, in accordance with Dr. Steven Siegel, chief psychological well being and wellness officer.
At Cedars-Sinai, Xaia — an acronym for eXtended-reality Artificially Clever Ally — was designed and programmed by Dr. Omer Liran, with concepts and analysis help from Dr. Brennan Spiegel and therapeutic enter from medical psychologist Robert Chernoff, in collaboration with the medical middle’s Know-how Ventures.
Dr. Omer Liran.
(Cedars-Sinai)
VRx Well being, a for-profit firm based by Liran, holds an unique license from Cedars-Sinai to market Xaia commercially. Cedars-Sinai and several other non-public traders maintain fairness within the firm.
A model just like the one Misty Williams makes use of is out there to the general public through the Apple Imaginative and prescient digital actuality headsets for $19.99 a month. A VR model for the Meta headset is freely obtainable solely to researchers. An online and cellular model is accessible to licensed clinicians for tiered pricing between $99 and $399-plus per 30 days, which permits them to ask sufferers to make use of the software.
Liran, a psychiatrist, mentioned Xaia is designed to complement, and never change, the providers of psychological well being therapists amid a nationwide scarcity of suppliers.
“Even if somebody needs to be seen once a week, they may only get seen once a month,” he mentioned.
The Xaia app attracts from lots of of remedy transcripts, each from actual periods and mock periods created by specialists to sound like an precise therapist.
For instance, if customers inform Xaia they’re combating a brand new most cancers prognosis, the robotic would possibly say, “That must be very hard for you,” then ask the way it’s affecting their temper, and what they discover themselves doing after they’re overwhelmed.
“Trying to stay positive when things feel so heavy must take a lot of energy,” the chatbot says. “When you notice yourself being pulled back into those difficult thoughts, what usually happens next? Do you find yourself withdrawing, or do you turn to anyone for support?”
Medical psychologist Robert Chernoff.
(Cedars-Sinai)
To this point, Xaia has been utilized by about 300 sufferers throughout varied analysis research at Cedars-Sinai, together with these targeted on persistent ache, alcohol use dysfunction, and irritable bowel syndrome, mentioned Spiegel, director of well being providers analysis at Cedars-Sinai.
Many individuals with persistent sicknesses additionally battle with nervousness or despair, Spiegel mentioned. Bodily and emotional signs feed off one another, and instruments like Xaia goal to assist with each.
The software isn’t coated by insurance coverage but, however billing codes for digital actuality remedy and digital well being providers do exist, and different hospitals just like the Mayo Clinic are starting to make use of them. VRx has an settlement to deploy Xaia at Mayo Clinic, in accordance with VRx Chief Govt Gabe Zetter.
Xaia isn’t the one app of its variety. Woebot, a pioneering chatbot developed by psychologist Alison Darcy whereas at Stanford, used scripted conversations primarily based on cognitive behavioral remedy to help customers with nervousness and despair.
Although it reached 1.5 million customers, the corporate shut down the app in July. Darcy mentioned the corporate is now targeted on constructing new instruments with massive language fashions, since AI is shifting quicker than regulators just like the Meals and Drug Administration can sustain.
Lately, some emotional help chatbots have been blamed for deepening misery, together with one incident through which a Florida teen died by suicide in 2024 after prolonged conversations with a chatbot.
Such incidents underscore the dangers of emotionally responsive AI instruments, mentioned Todd Essig, a psychologist and founder and co-chair of the American Psychoanalytic Assn.’s Council on Synthetic Intelligence.
“Even after the most loving, empathic response, an AI doesn’t care if you drive to the store or drive off a cliff,” Essig mentioned.
AI packages study to imitate human responses, Essig mentioned, so it’s as much as the folks constructing them to set clear limits and guarantee they don’t trigger hurt.
“Mentally and physically, I feel more at peace,” Misty Williams mentioned after a session with AI-powered Xaia.
(Genaro Molina / Los Angeles Instances)
When constructed with moral frameworks and used underneath medical supervision, instruments like Xaia can help real therapeutic progress, functioning extra like digital journals — a contemporary twist on the paper workbooks given to sufferers many years in the past, mentioned Jodi Halpern, a professor of bioethics and medical humanities on the UC Berkeley Faculty of Public Well being.
However many emotional help chatbots that aren’t clinically monitored are designed to imitate intimacy and construct emotional bonds.
“People can experience the app as another,” Halpern mentioned. “But it’s not actually giving them real-life experiences with other humans that are important for developing the healthy, mutually empathic curiosity that people need to participate in complex human relationships.”
Halpern famous there’s a distinction between clinically accredited psychological well being instruments and people with no oversight. She and others are supporting a California invoice sponsored by state Sen. Steve Padilla (D-Chula Vista) that may require firms creating psychological well being chatbots or apps to reveal whether or not their instruments are clinically validated, regulated by the FDA or depend on generative AI.
Liran mentioned he and his companions are conscious of the constraints and have inbuilt guardrails to maintain the chatbot from saying something dangerous or inappropriate. For instance, one arm of the AI generates the response, and one other immediately double-checks it to ensure it’s secure earlier than letting it by means of to the consumer.
“We’re not just opening it up to the public,” Liran mentioned, declaring that the guided remedy model on cellular and desktop is out there solely by means of a licensed clinician proper now and Cedars is testing Xaia in a number of research. “We’re trying to be very careful.”
In a 14-person examine, sufferers utilizing Xaia with delicate or average nervousness or despair opened up about quite a lot of subjects, together with a mom who handed away and worry of being laid off. For a affected person who had been having night time sweats since a breakup, Xaia requested to listen to extra about what made the connection really feel unresolved and the way it affected the affected person.
A few of the sufferers nonetheless most well-liked the nuance and responsiveness of a human therapist, however the medical literature suggests sufferers are warming as much as the thought of a nonhuman therapist.
In a examine revealed in PLOS Psychological Well being in February, individuals had been requested to check responses written by licensed therapists and people generated by ChatGPT.
Not solely did many battle to inform the distinction, they persistently rated the AI’s replies as extra empathic, culturally delicate and emotionally partaking.
Xaia’s creators see the software as an extension of the patient-therapist relationship. It’s the sort of factor that is perhaps helpful if somebody wants psychological well being help in the course of the night time or between periods.
“We still need therapists — humans — to look other humans in the eye to have conversations about vulnerable topics,” Spiegel mentioned.
On the identical time, “it’s not practical to simply bury our head in the sand and say we shouldn’t do this, because AI is everywhere,” he mentioned. “We’ll be brushing our teeth with AI before long.”
