AI Hype vs. Reality: A Conversation With Nuance (Microsoft) CSO, Peter Durlach

Peter Durlach grew up around computers and was exposed to
innovative technology at an early age by his father, who helped
run a lab in the electronics department at MIT. After graduating
college in the 1980s, Peter caught the entrepreneurial bug way
before it was cool, and was employee number four at a company
developing the first voice user interface for the
Macintosh.

Peter’s work put him at the center of bringing voice recognition
and artificial intelligence (AI) to healthcare – work he
continues today as EVP and Chief Strategy officer at Nuance
Communications, a Microsoft Company.

At Articulate Systems, the first voice recognition company Peter
helped build, he and his team eventually recognized that 60% of
their user base was physicians dictating medical notes. He helped
pivot the company to focus exclusively on healthcare and build
the technology that became PowerScribe, a solution still used by
most radiologist today.

Through acquisitions, Articulate Systems eventually became
Nuance. After some time away from the company as a software
consultant and running an AI-powered contact center business he
sold to Microsoft, Peter was recruited back to Nuance in 2006 to
create its healthcare division. By 2019, Nuance was focused on
the healthcare and customer engagement market, and in 2022 was
acquired by Microsoft for $20 billion.

In this episode of Healthcare is Hard, Keith Figlioli builds on
earlier conversations with guests like Mayo’s John Halamka and
Advocate’s Rasu Shrestha to unpack the hype around AI in
healthcare and understand what’s real right now, and what will be
in the future. Some of the issue Keith and Peter discussed
include:



The AI adoption curve. Peter described the different
adoption curves he sees for different use cases of AI in
healthcare. He talks about how adoption of administrative use
cases will happen much faster than clinical applications, and
the factors that will influence adoption curves – from
performance requirements to governance and patient safety.


Use cases that matter. Nuance and Microsoft see one
common theme for the application of AI in healthcare – that it
should be a copilot and not an autopilot. In the near term, the
major focus will be on automating administrative tasks in
revenue cycle, payment integrity, documentation or other areas
where there is a large labor expense. While automation may be
able to accomplish 30% of the work in some areas, and 70% or
more in others, the goal is reducing labor while maintaining
human oversight. The bar is much higher for clinical use cases,
but the same rules apply.


Opportunities for startups. Peter shares lessons from
his time at startups and the world’s largest technology
companies. He talks about how incumbents have an unfair
advantage and a “right to win” because of their footprint with
trusted clients, and how speed and nimbleness give startups an
advantage. He says startups focusing on AI need to understand
if they’re supercharging an existing process, or creating a
truly new paradigm, because understanding this will dictate
their path to success.


Governance – Peter talks about the governance and
regulatory issues impacting AI adoption and the potential for
certifying technology for specific uses, similar to what’s
required for pharmaceuticals or software as a medical device.
He also discussed how Microsoft is building safeguards within
its cloud infrastructure, allowing the ecosystem to stay
focused on innovation.



To hear Keith and Peter talk about these topics and more, listen
to this episode of Healthcare is Hard: A Podcast for Insiders