#Powered by Talent - Niels Olson

Niels Olson is a board-certified pathologist and the Chief Medical Officer of the Defense Innovation Unit in Mountain View, California. His previous duty stations include Laboratory Medical Director at Naval Hospital Guam, staff pathologist at Naval Hospital Camp Pendleton, and General Medical Officer for Surface Forces Pacific. He completed AP/CP residency at Naval Medical Center San Diego. His undergraduate degree is in Physics and prior to medical school he served 7 years as a Surface Warfare Officer in the U.S. Navy. He oversees research programs in machine learning for a broad spectrum of anatomic pathology applications, augmented reality microscopy, artificial intelligence applications in radiology, and distributed sensors in austere environments.

As a part of the Human Systems Portfolio, Niels’s expertise has been essential to the development of Predictive Health and Rapid Analysis of Threat Exposure (RATE) solutions. To read more about these projects, please see our 2020 Annual Report.

Interview

Why were you interested in working at DIU?

I thought the mission was spot-on, addressing the glaring lack of modernization in the DoD and the value of leveraging the R&D done in the commercial sector.

What is your favorite DIU experience?

Working with incredibly talented people from a variety of backgrounds (commercial and operational) and from each Service, every day. I’m learning from everybody all the time.

If you could solve any DoD problem tomorrow, no matter how big, what would you tackle and why?

I’d tear down the computer systems and start over with cloud-based everything, run by a central organization responsible for their own code, their own infrastructure, their own networks. It’s so heavily Balkanized that we’ve become our own worst enemy.

What emerging commercial technologies are you most excited about?

I think AI is going to take over a lot of things, so excited, but nervous.

What are you reading right now?

Karl Popper, The Logic of Scientific Discovery. It’s a landmark work, which established falsifiability as a central characteristic of science. This actually presents a challenge for scientists in the real world, because virtually any statement a scientist makes can be falsified if one looks hard enough for alternative statements. For better or worse, this leaves an opening for people who might not like some facts that the vast majority of reasonable scientists may readily agree on. This is important because that opening has been used with significant effect, both to drive enormous advances the community once thought impossible and fan the harmful flames of anti-science sentiment in modern society.

What defense challenge/commercial solution are you working on right now?

Predictive Health, which has a few components: two in pathology and one in radiology. In pathology, we are developing 2 new machine learning models, which is a huge lift. We’re also deploying augmented reality microscopes to VA and Military hospitals so pathologists can test them. In radiology, we are deploying multiple machine learning modules for radiologists responsible for diagnosing chest x-rays and CT scans.

Military: what do you wish DIU could have solved for you in the past?

Fix the DoD’s computational and network infrastructure. The idea that I need to type identifying information into a form in 2021 is insane. They should know who I am. They have my DNA. Timesharing computing was a solved problem in the mid-1970s. A war involving AI is going to involve machines that move so fast they will be able to adjust fire as a bullet moves through the barrel of the gun. Ignoring cyberwar, nuclear war, how are we going to win that kind of war, machine-enabled, high-velocity war, when we can’t efficiently identify people and flow data among databases?