(Franklin Manchester, Global Insurance Strategic Advisor, SAS. Photo by author.)

For SAS (Cary, N.C.), ITC 2025 is more than an opportunity to meet customers—it’s a venue to advance an urgent conversation about AI trust and governance. “Even though it’s a U.S.-based event, we’ve been impressed with how delegations from around the world come here and participate,” comments Franklin Manchester, Global Insurance Strategic Advisor, SAS. “But what we’re noticing of increasing importance is a conversation around AI governance, around trust.”

That concern, he says, is rooted in findings from a recent IDC report commissioned by SAS, Data and AI Impact Report: The Trust Imperative. “Respondents trust generative AI 200 percent more than they trust machine learning,” he notes. “That doesn’t make any sense because machine learning has been around for 20 or 25 years. The issue is that people don’t know what it is—they’re more comfortable with generative tools because of the sentiment associated with them.”

The Perils of Misplaced Trust

Manchester argues that this imbalance underscores a dangerous tendency to overestimate emerging technologies while overlooking the fundamentals of data integrity and governance. He recounts a recent example in which a consulting firm’s report for a national government—produced partly with Microsoft Copilot—contained fabricated citations and non-existent authors. “That’s reckless and foolish,” he says. “I don’t believe generative AI is making us dumb; I think it’s making us careless.”

For an industry built on reliability, he adds, that kind of carelessness is unacceptable. “If you’re a consultant or an auditor, your whole existence is based on trust. You can’t mess around with that.” The SAS report, he continues, shows that while many organizations are investing in AI, “the majority of respondents aren’t investing in the trust aspect just yet.” The challenge, he says, is to develop governance before giving AI greater autonomy: “I’m interested in them getting there before they give agentic AI agency—forgive the pun—to do certain things that create significant downside risk.”

Manchester cites studies showing that bias is already embedded in large language models, including evidence from Lehigh University that generative AI systems have denied loans to minority applicants while approving equivalent ones for white applicants. “So, from SAS’s perspective, we want to talk about AI governance, we want to talk about data integrity,” he says. “If you’re going to build a model, let’s figure out how you govern it so that when it drifts or hallucinates, you can take accountability.”

Shifting from ‘Human in the Loop’ to ‘AI in the Loop’

Manchester acknowledges that perfection isn’t the goal. “If a human performs at 95 percent quality, you’re thrilled,” he says. “But if a machine makes one single mistake, people champion that error and say, look what this dumb machine did. The expectations are unrealistic. You’ve got to design for trust.”

He sees encouraging signs in how business leaders are beginning to reframe the relationship between humans and AI. “There’s a shift from ‘human in the loop’ to ‘AI in the loop,’” he explains. “Figure out where you want to use AI—and keep your people doing what they do best.” Rather than viewing AI as a job threat, he says, insurers should see it as a tool for capacity creation. “Most of the customers I speak with are already overworked,” he notes. “They’re in meetings four to six hours a day, working after hours, answering 150 emails. We need AI to create capacity for ourselves to breathe.”

That shift in thinking, he says, should move beyond productivity to sustainability. “Can I do what I need to do in a 35-hour work week instead of 60?” he asks. “Once we create that capacity, we can reinvest it in workforce development—helping the people who know the business learn how to use the tools.”

Staying Grounded in Strategy

Ultimately, Manchester believes insurers should resist the hype of “AI strategies” in favor of pragmatic business alignment. “We didn’t have a Windows strategy in 1995 or a Google strategy in 2006,” he says. “Those were tools we learned how to adopt. If we stick to our business strategy and figure out where AI fits in, we’ll succeed.”

That pragmatic outlook defines SAS’s approach to AI. “We’re not chasing the next shiny thing,” Manchester concludes. “We’re helping insurers think critically about how to use AI responsibly—governing it, measuring it, and making sure it serves the people who depend on them.”

ITC Briefing: Percipience’s Bruce F. Broussard, Jr. on Data for AI

Socotra CEO Dan Woods on Innovation and Insurance’s Technology Gap

ITC Briefing: EY’s Chris Raimondo on Insurers’ Enterprise Turn Toward AI

ITC Briefing: Fenris’ CEO on Predictive AI and Distribution Efficiency