Eight months after testifying in Congress about the use of artificial intelligence (AI) as a tool to combat waste and fraud in the federal government, a watchdog agency’s lead scientist said that idea is a long way off and he’s still waiting for a response from lawmakers.
Sterling Thomas, chief scientist within the Government Accountability Office (GAO), leads the watchdog agency’s research surrounding AI. He told The Epoch Times that he has advised lawmakers that the implementation of AI could drastically decrease fraud and improper payments, but that it faces serious barriers to entry.
Thomas testified on Capitol Hill in April, giving recommendations on what it will take to use AI effectively within the federal government. Since then, he said, Congress has not adopted his recommendations nor directed him or his team to continue work on this plan.
“We haven’t done another study, and we’re waiting for Congress to basically tell us to go back and evaluate this because they’re the ones who direct our work,” Thomas said.
One of the recommendations Thomas gave lawmakers earlier this year to fight fraud and improper payments was to begin planning on how to develop an AI-skilled workforce, which he said is crucial.
To his knowledge, Thomas said, no movement has been made in this direction.
AI could potentially identify indicators of potential fraud, but an employee would need to be the one to make a final determination, the chief scientist said. This same employee would need to train in anti-fraud, understand indicators provided via the AI tool, work with those tools, and investigate potential fraud, all while having an understanding of the differences between improper payments, waste, and abuse.
“If you don’t bring in the workforce, but you just bring in the tools, they’re not going to work the way you want them to,” Thomas said.
In addition to training skilled employees, the AI itself needs to be trained correctly, he said, meaning a significant amount of error-free, unadulterated data is needed.
“I have no doubt, in the administration, there are folks who are trying to improve the quality of data. … It’s not a secret that the quality of data has a major impact in how these AI tools work, but it’s a big job. So, it’s one of those things that is going to take a significant amount of time as well as money,” Thomas said.
The GAO scientist said his office never worked with the Department of Government Efficiency (DOGE), an agency with a similar goal in rooting out government waste, outside of sharing already publicly available information. Thomas said Congress has tasked his office with evaluating DOGE’s approach with a study to be published at a later date.
Thomas said he is enthusiastic about AI’s use in the future in government but is cautious about moving too fast.
