Artificial intelligence dominated many conversations at this year’s BenefitsPRO Broker Expo in Chicago. In Tuesday’s session, “The AI Maturity Spectrum,” speakers discussed the art of AI implementation and how it can make a positive difference in the way the benefits industry works.
Rather than focusing on specific tools or simply stating how AI can do our jobs for us, Ben Conner, CEO at Conner Insurance and Jason Beutler, Founder and CEO at Robosource, walked through the journey of how AI should be implemented through a maturity journey that Beutler titled “The AI-Maturity Spectrum,” which consists of 4 phases:
- Vibing
- AI-Assisted
- Agentic Workflows
- Multi-Agent Workflows
The session described the power of viewing AI implementation as a spectrum that is slowly implemented as a company learns more of their needs; the tools are partners rather than robots that are replacing human jobs.
See also: AI in HR strategy: An expert’s 5 pillars for successful implementation
“We move up this spectrum until we end up with this mythical thing that you’ve all seen over the news: multi-agent workflows,” Beutler said. “This idea that somebody has their entire business running on AI, and they don’t have to touch anything.”
“For the most part, that’s not real,” Beutler said.
Phase 1: Vibing
During the vibing phase, AI is used as a tool similar to Google. Employees are asking the tool questions, and it’s spinning out the answers they need quickly. The biggest problem this phase sees is hallucinations. Context matters, which takes the company into the next phase.
Phase 2: AI-assisted
The key to this phase is providing the tools with the right context to perform tasks accurately and save the company’s time.
The tools need a role or persona, clear instructions on workflows, background and finally, constraints and audience details.
“By giving the AI tool structure, it’s going to give you much, much stronger output,” Beutler said.
Phase 3: Agentic workflow
The third phase, agentic workflow, is really about trusting the AI tool to own the task and output. Companies need to master the first phases before moving into this one.
Phase 4: Multi-agent workflow
The last phase of the spectrum is what is often reported on as how AI can replace humans. But to get to this phase, the AI tool would have to delegate the work, understand the context in which the company needs the work done, and then successful 100% of the time to produce accurate, clean results.
“This is not happening,” Beutler said. “Because if you’ve ever played with the tools, they’re really, really hard to use.”
While AI is changing the way we work and is already useful, the way it is implemented makes or breaks the value it can bring to an organization.
“AI can allow our people to not be bogged down in the junk and to allow them to invest in the client experience,” Conner said.
Credit: Source link
