The National Science Foundation's Small Business Innovation Research program has awarded Actualization.AI, an artificial intelligence startup, a $275,000 grant.
The grant will support the company’s efforts to develop AI solutions that strictly adhere to human rules, ethical standards, and legal guidelines.
“We’re concerned with the lack of tools to ensure AI is doing what humans tell it to," John Licato, Actualization.AI’s founder and owner, said in a statement. “We are on a mission to fix that, so we can help cure the world’s 'AI anxiety.’”
Licato is an assistant professor at USF's computer science and engineering department. He's also the founding director of the college's Advancing Machine and Human Reasoning Lab, a cross-disciplinary lab studying ways to improve the human reasoning of AI.
Licato founded the lab in 2017. The company was founded in 2023 after the realization that the two entities' missions both focus on improving reason in people and AI, and a commercial application was viable.
Following human rules
The NSF grant focuses on commercial impact research, providing a project framework for AI systems that follow human rules, especially in laws, policies, or contractual agreements.
Because today’s societies distrust interactive AI agents and chatbots, Licato said this is an important area of AI research.
“Most industries have adopted these aids, despite their known limitations, particularly [given] problems of hallucination[s] and [the] inability to behave in accordance with the given policies,” Licato said in a statement.
One of Actualization.AI’s key projects involves developing a secure, streamlined algorithm to convert complex rules, laws, and guidelines into formats AI systems can better understand. Licato said the process could be applied in various industries, including finance, cybersecurity, customer service and medical insurance claims.
“Given that all industries with a customer interaction component are turning to chatbots, the economic impact of the project is significant,” Licato said in a statement. “Our work accelerates the scientific and technological understanding of rule design, so systems are better interpreted by different humans as well as artificial intelligence systems.”
