Artificial intelligence and GPT dominated chatter at HIMSS, but not all AI is created equal, says Suki AI CEO Punit Soni, Axios' Claire Rychlewski writes.

Why it matters: Cutting through the noise is important, especially in a risk-averse market.

What he's saying: Soni, who previously worked on the development of Google Assistant during his tenure at the tech giant, says AI in health care is only as useful as the surrounding infrastructure.

• Further, some tech companies purporting to offer AI capabilities may not have a pure software solution, he adds.
• "The first and easiest way to figure out if someone is actually doing AI is asking the price and the turnaround time," says Soni.
• "I can tell you what it shouldn’t cost — a couple thousand dollars or more per user, per month," he says. "And software doesn't take four hours. Software is instantaneous or it's not."

The intrigue: Seasoned investors don't want to invest in AI models without clinical use cases or health care infrastructure, says Lucem Health CEO Sean Cassidy.

• "There’ll be consolidation," he predicts.
• "There is competition out there. There’s more than one AFib detection algorithm. The question is, what's the differentiation between those models?"