by Dr. Elisa Borah
Trying to keep up with the ever-expanding AI landscape can feel overwhelming. It’s hard to stay on top of all the advances and new apps, or to know whether the information we consume is AI-generated. As AI becomes ubiquitous in our personal lives and at work, it’s our responsibility to understand and harness it to benefit us and those we care about. As leaders, we need to identify how AI supports our goals and values and routinely assess and validate its outputs to deliver the best answers and outcomes. At the same time, we also must embrace its immense potential to achieve results we never imagined.
A constant theme among social work scholars grappling with this new world is cautious optimism—caution regarding the power and allure of AI, mixed with excitement about the magnitude of problem-solving we can achieve to improve the speed of our research and practice, e.g., improving our caseload risk analysis to address mental health crises or understand better which factors influence substance abuse recovery, is a dream for those of us who became social scientists to help people through research. In social work, we must assess the quality of the data on which AI models are trained, identify when they do identify when it does not reflect our profession’s values or an equitable society, and build new models that do. We need to work together to design AI tools using unbiased, truly representative data so we can accelerate our ability to address the needs of all clients and communities. Social workers are uniquely positioned, through our training and code of ethics, to inform AI solutions designed with clients’ confidentiality, delivery preferences, and outcomes in mind. Very few tools like this exist for social work, although there are early products, such as Bonterra’s Que and Social Work Magic, that were designed with service-delivery experts. A new era of immense innovation is underway, and we need to embrace it to define how AI impacts practice and client outcomes. This work will require teams with diverse expertise: social workers across practice areas, computer and information scientists, community-based service organizations, policymakers, and industry product development teams to create the best solutions.
Beyond service-delivery innovation, there is a dire need for ‘living’ policies to guide how we train our students and maintain best practices for clinical and administrative work. Because of AI’s rapid evolution, organizations at all levels need new approaches in policy creation and maintenance. This type of guidance is a common need voiced by social workers and organizational leaders. A majority (73%) of the 860 social workers surveyed in our study, through the National Association of Social Workers (NASW), believe AI will change the future of social work. They also indicated many areas they require support: understanding AI’s impact on bias, vulnerability, and safety (64.2%); developing ethical guidelines (62.3%); developing improved privacy protections (54.3%), and training on AI tools (51.3%).
In this time of rapid change and immense opportunity, we need to keep key questions in mind:
- How do social workers across diverse practice fields develop decision-making around which types of work we allow AI to do, where it’s safe to rely on it exclusively, and which tasks should be done by humans alone?
- How do we include stakeholders across social work (e.g., service delivery, education, research) in the design of ethical AI models and policies and policy frameworks that they can adapt to their local needs and practice environments?
- How do we work to design AI to harness the best of humanity’s collective knowledge, not just what available data it was trained on?
Ultimately, the future of social work practice and research aimed at improving our communities’ well-being is up to us, in how AI is used and to what end. It’s our job to lead this work, not be latecomers to a technological sea change that we will later have no power to change. It’s our job to invite the right people to the table to create the best tools that meet our ethical standards.
(In case you are wondering, I did not use AI for any part of this piece! It was tempting, but I’m trying to keep my own cognitive juices intact.)
