Keeping the “Human” in Human Services
For mental health practitioners and peer support specialists, artificial intelligence poses both benefits and risks, warns a Rutgers professor of behavioral health care
In behavioral health, lived experience is an essential asset for peer supporters – those who use their own experiences to help people feel empowered, supported and hopeful.
In a new policy paper published in Psychiatric Services, Margaret “Peggy” Swarbrick, associate director of the Center of Alcohol and Substance Use Studies at the Rutgers Graduate School of Applied and Professional Psychology, and Nev Jones, an associate professor of social work at the University of Pittsburgh, warn of the risks when providers lean too heavily on robots to offer peer support.
“AI has the potential to increase access to needed care, reduce the administrative burden of peer supporters and augment peer support interactions,” they write, adding: “Without careful implementation, AI may compromise the relational qualities, privacy and ethical principles that underpin peer support.”
Swarbrick expands on these findings and urges mental health specialists to keep the “human” in human services.
How are peer supporters in behavioral health using artificial intelligence in their work?
AI often appears in the form of chatbots and other digital tools. Peer supporters use AI to help clients navigate a problem or search for resources, like finding a food pantry or accessing affordable housing. Peer supporters can also get interesting ideas from these tools – like brainstorming the development of materials for individual and group meetings.
Beyond these examples, AI has the potential to increase access to needed resources and support, reduce the administrative burden of peer supporters, and augment peer support interactions.
But AI-driven tools are also expanding into more questionable areas. Standalone peer support agents have emerged, and researchers in Korea have even tested a chatbot that portrays itself as someone who has recovered from an eating disorder.
These tools don’t possess human lived experience or empathy, and don’t have the capacity for ethical judgment. Thus, in peer support contexts, it’s important to distinguish AI-assisted but human-led peer support from standalone AI “peer support” agents.
What are you doing to help providers navigate these nuances?
Part of it is simply raising the issue. Our goal with this paper is to help educate peer supporters about the limits of AI, but also to illustrate the ways it can be helpful if used properly.
We are also developing guidance on how peer supporters can use AI responsibly, including how to write effective prompts, verify information and protect privacy. We are exploring how they are using AI in their peer support work and offering a framework to ensure that peer support values such as mutuality and empowerment remain central.
Is this a ubiquitous problem, or is it isolated?
This concern extends beyond peer support. Similar warnings are emerging in fields like counseling, education and health care, where practitioners worry that convenience tools could replace, rather than support, human judgment and presence. The danger isn't AI itself; it's the potential for overreliance to replace the human element that these fields depend on.
What are the dangers of overusing AI in behavioral health?
Patient privacy is one. AI hallucinations – when chatbots make up answers – and AI-driven misinformation and bias are others. Although the use of AI tools by peer support professionals, rather than directly by users, may insulate people they support from some of the biggest direct risks of chatbot interaction, peer supporters could still encounter hallucinated resources or other difficulties related to technical assistance. It’s essential that these tools be used sparingly, and with great care.
Simply put, we must protect the human piece in peer support. Human connection is what makes peer support work. That connection cannot be automated, streamlined or replaced.
At the end of your paper, you discuss regulatory frameworks and policies that should be considered. What are you recommending?
We recommend several professional and regulatory improvements to protect patients and safely integrate AI into peer support practice.
First, AI training should focus not only on technical proficiency but also on ethical considerations, alignment with peer support values and decision-making in relational and crisis contexts.
Second, organizational policies should clearly define AI’s supportive role, outline procedures in the event of crisis escalation following AI feedback and ensure adherence to privacy standards.
But the most important thing we can do is share our experiences. As peer-support providers, we must keep educating one another about AI’s benefits and risks.
