Page Anderson, PhD, is bullish on what digital mental health can do for the field and the world

More than a quarter of practicing psychologists worry that artificial intelligence may make some or all their job duties obsolete, according to APA’s December 2024 Practitioner Pulse Survey. But Page Anderson, PhD, ABPP, an associate professor of psychology and neuroscience at Georgia State University, is not one of them.

Anderson, also a licensed clinical psychologist with a board certification in behavioral and cognitive psychology, doesn’t consider herself a technology person but is bullish on what digital mental health can do for the field and the world. She studies acceptance and use of digital interventions by both clients and clinicians and sits on the board of directors for the International Society for Research on Internet Interventions.

At the same time, Anderson acknowledges that psychologists face challenges around adding tech to their work, including how to do so in ethical and equitable ways. She covers these and other considerations in her APA continuing education course, “The Chatbot Cannot Replace You: Using Digital Technologies to Expand Your Positive Impact on Mental Health,” and shared insights with the Monitor on how psychologists can thoughtfully integrate such tools into their practice.

A recent workforce analysis by the Substance Abuse and Mental Health Services Administrationopens in new window estimated the number of additional behavioral health providers needed to meet the current demand in the United States for adults with substance use and serious mental illness—that number is 4,486,865. There’s essentially no way that we can meet the mental health needs of our counties, states, country, or the entire world with our current model for providing mental health services.

Many psychologists get into this field because we want to help people, have good interpersonal skills, and like to connect with others. Tech may not be what we like to lean on, but it’s an extremely powerful way to help people. We’ll always have a big need for in-person services, but we need more than that.

There are two main barriers to people accessing and using mental health services: practical barriers, such as cost and getting time off work, and attitudinal barriers—including mental health stigma, lack of perceived need, and our desire to handle issues on our own. Technology can address the former, but without a clear strategy, it won’t necessarily solve the latter.

For example, research from my lab shows that people who have more barriers to accessing in-person services also show more skepticism toward internet-based cognitive behavioral therapy. That’s a very troubling finding.

Building a digital mental health service doesn’t guarantee people will use it—or use it long enough to benefit. Dropout is a huge problem. The earliest digital programs repackaged empirically validated in-person treatment for anxiety and mood disorders to deliver on the internet, like a digital self-help book. But that’s not how we use technology. We must design digital interventions that fit with how we use technology in the real world.

People’s attitudes toward artificial intelligence seem better. It could be that people feel engaged, they have self-efficacy, they design their own prompts, and they can get immediate feedback that feels personalized. That’s much different from a 12-week digital mental health intervention for depression that’s been packaged, tested, and approved.

Ask yourself: What would help me? If you’re a clinician, and you want to know how somebody’s sleep is, you could rely on self-report or you can ask them if they’re willing to share data from their Fitbit. If you’re tired of repeating psychoeducation about cognitive behavior therapy, you can explore ways to reinforce that message with technology, such as sharing a relevant YouTube video. And if writing progress notes is time-consuming, artificial intelligence could be used to help you.

Access continues to be an issue. Many of us take broadband access for granted, but there are still many people in this country without it. There’s also growing concern about a new sort of digital equity gap—that those who “have” will be treated by people, while those who “have not” will be treated by algorithms.

Not only digital access, but digital literacy is also important. Our population is rapidly aging, and technology is one way to serve that group. But not if people don’t know how to use it. Training on new technology is the kind of thing we need to be building an infrastructure for now. For example, college graduates with undergraduate psychology majors could serve as digital navigators, working closely with people using digital technology to ensure they have the skills and confidence to use it.

Another real concern: One of the main things that promotes equity is transparency, but most AI algorithms are proprietary, business-owned algorithms. That’s a problem first because we don’t know how the model is trained. Most of them are trained on publicly available data, and all of humanity is in that data—so, the good, the bad, and the biases baked into the data will also be baked into the algorithm. Another problem is that algorithms are typically developed without input from clinicians with on-the-ground expertise, which can result in the use of proxy variables that fuel inequities. That’s another way that bias gets in, but it’s often invisible.

Oftentimes, the approach to developing new technology is: Here’s a solution; let’s find a problem it can fix. Using technology to increase equity in mental health services begins with a very careful framing of the problem. To do that, we’ll need to work with community groups, clinicians, and the people who could benefit from services but for whatever reason, cannot or will not access them.

We have the opportunity to build technology equitably from the ground up, versus what has happened with in-person services, where a model was built first and equity was considered later on. We have a chance to do it differently through technology, but it won’t happen unless it is very intentional.

I was awarded a training fellowship designed for clinicians from AIM-AHEAD, a consortium funded by the National Institutes of Health to ensure better health for all through artificial intelligence and machine learning research. I’ll be developing an AI algorithm based on social determinants of health that identifies people and places most likely to benefit from technology-based interventions. I’m also developing a doctoral-level course and continuing education workshops to teach clinicians about how AI works and the ethical and equity issues related to its use, as well as a social media strategy to educate the public about these considerations around AI and mental health.