Executives and researchers from Mizuho Americas, AMD, flipando.ai and Credo AI sat down to talk about what advice they have for companies, as well as the success stories and risks they’ve witnessed in the early days of generative A.I., during a special panel at Fortune Brainstorm Tech this week.

One common refrain was the importance of using the technology responsibly and competitively.

“Responsible A.I.is not just like an ‘Oh, a nice to have’ thing. It’s becoming actually a critical component of businesses,” said Navrina Singh, the founder and CEO of Credo AI, which performs risk and governance analysis of A.I.

She pointed to an insurance group that used Credo to create an impact report detailing how to prevent racial discrimination in life insurance policy offerings. It increased the company’s customer retention and brought them new customers, Singh said. 

For filipando.ai CEO and founder Maia Brenner, helping A.I.-curious companies “walk, not run” has been the key to success. Brenner’s startup helps other companies build generative A.I. apps. Despite the crowded field of giant vendors selling A.I. products, such as OpenAI, Adobe, and Microsoft, Brenner has managed to carve out a niche.

The success of generative A.I. shouldn’t be measured simply by how commonplace it becomes within the business landscape; the real success stories, many of the panelists noted, will be the companies that use the technology in a smart, purposeful manner rather than just jumping on the bandwagon.

Credo’s Singh said that the most A.I.-savvy companies are “being very cautious in terms of creating the right sandboxes with the right safeguards and controls to make sure that as they’re testing these applications, in very high-value areas that they do so with caution and with the right guardrails in place.”

Still, for all the responsible and meaningful A.I. applications, there’s also going to be the “roadkill.”

“When you think about the automation and the productivity benefits that Gen A.I. can unlock, I think that’s at least somewhat problematic for RPA (robotic process automation),” said Gregg Moskowitz, a senior enterprise software equity research analyst at Mizuho Americas. 

It’s still early, but “there are puts and takes, there are offsets and we have to sort of see how all that plays out,” Moskowitz said in reference to which industry, people or companies could be knocked out by generative A.I.

The competition extends beyond the makers of large language models and the existing businesses they threaten to replace. Hardware has been among the most fiercely competitive A.I. battlefields, with chipmakers at the center of the action. According to Mark Papermaster, executive vice president and CTO of AMD, the competition is spurring innovation.

“We’ve really optimized that and we’re trying to look at really, understand with our customers, the workloads to fine-tune these, and not only provide competition but as I said, I really think about sustainability. We set specific goals and markers in that regard,” Papermaster said.