Benefits Innovation & Access

Although I often take pride in my resourcefulness, it – and my pride – were acutely put to the test after I got myself into a $46,000 emergency surgery, only to find out I was uninsured. I began calling phone numbers and bookmarking links before I was even discharged from the hospital. But it still took six months of delayed response time, inaccurate information, and occasional professional condescension before my bill was resolved through Medicaid. This experience motivated me to work at a nonprofit call center for over a year because I wanted to support others with the expertise and empathy that I wish I had received when applying for Medicaid.

The organization I worked for unfortunately shut down due to financial distress, and too much attention has been paid to the call center’s allegedly high-contact, low-efficiency customer service methods without consideration of the unique and adverse environment that influences public benefit clients. 

Someone who is not familiar with the environment of benefits enrollment may wonder why it costs more than $283 to help someone enroll in food stamps, particularly when they need or already want the assistance. Those with knowledge of the system would know the enormous barriers in receiving, recertifying, and reapplying for even the most experienced users. The call center specialist, for example, would be crucial for a client who has never heard about a process like DHS appeals but who intuits that a benefit is being rejected due to a clerical error. The call center specialist’s expertise is superb tacit knowledge, or knowledge that can be executed but not explicitly explained  – and until recently, not even remotely capable of execution by artificial intelligence (AI).

How might this inform service systems for change that are in similar situations? First, there are different ways to measure efficiency. Call center assistance may seem too expensive when AI models can do it for three percent of the cost, but how many hard-to-reach clients are more likely to use the former method to reach out for help than the latter? How many clients relied on specialized, personal assistance to break the cycle of Medicaid churn and achieve a continuity of care that many take for granted? Asking different questions may not only make a difference to funders, but also focus innovation and product development on service specialists, whose work can be aided with AI without being replaced. 

Second, technological limitations are not mere conceptual challenges but rather non-negotiables to which product design must adapt. For example, large language models like ChatGPT are currently maximized through text input. Call center specialists provide support through the phone, which can be easier for those without immediate access to a computer, those who are busy caregiving, and all those who feel higher amounts of emotional distress than even those experiencing food insecurity without government support. And alternative products driven by AI or client self-service may lose impact on such populations.

Although I have forgotten many names of the resources I accessed, organizations I valued, and even some of the great people I spoke to when I applied for Medicaid, what I remember vividly is the information I received from receptionists, intake workers, and knowledgeable friends that was accurate, empathetic, and respectful of my ability to make my own decisions  – those people’s words made me feel I wasn’t alone in an informational black box. 

To truly innovate in the field of public benefits, administrators, funders, and policymakers should prioritize the integration of empathetic, knowledgeable human support with AI tools to ensure that those who need support the most are not left behind in the name of efficiency. By focusing innovation on existing human relationships and leveraging AI to complement the expertise of service professionals, we can create a future where public benefits are accessible, equitable, and efficient.