In this piece, the Digital Benefits Network shares several sources—from journalistic pieces, to reports and academic articles—we’ve found useful and interesting in our reading on automation and artificial intelligence.
This publication seeks to answer one of the most common questions that CIOs ask: “What are other states doing with generative AI and what is the role of the state CIO?”
National Association of State Chief Information Officers (NASCIO)
Takeaways from a workshop focusing on applying human-centered design to government artificial intelligence (AI) projects, led by Elham Ali, Researcher from the Beeck Center for Social Impact and Innovation.
The state of South Dakota Bureau of Information and Telecommunications (BIT) designed guidelines for the responsible use of AI-generated content in state government agencies, emphasizing the need for proofing, editing, fact-checking, and using AI-generated content as a starting point, not the finished product.
South Dakota Bureau of Information and Telecommunications
The team explored using LLMs to interpret the Program Operations Manual System (POMS) into plain language logic models and flowcharts as educational resources for SSI and SSDI eligibility, benchmarking LLMs in RAG methods for reliability in answering queries and providing useful instructions to users.
NYC's My File NYC and New Jersey's unemployment insurance system improvements demonstrate how successful digital innovations can be scaled across various programs, leveraging trust-building, open-source technology, and strategic partnerships.
The team introduced an AI assistant for benefits navigators to streamline the process and improve outcomes by quickly assessing client eligibility for benefits programs.
An interactive chatbot that helps SNAP participants and the public ask questions and receive guidance about SNAP work and community engagement requirements in conversational form.
The team developed an AI solution to assist benefit navigators with in-the-moment program information, finding that while LLMs are useful for summarizing and interpreting text, they are not ideal for implementing strict formulas like benefit calculations, but can accelerate the eligibility process by leveraging their strengths in general tasks.
This academic paper examines predictive optimization, a category of decision-making algorithms that use machine learning (ML) to predict future outcomes of interest about individuals. Through this examination, the authors explore how predictive optimization can raise concerns that make its use illegitimate and challenge claims about predictive optimization's accuracy, efficiency, and fairness.
