Thinking differently, driving innovation at DARPA

Board Leadership Center (BLC): “Preventing technological surprises”—DARPA’s mission—is something that keeps many CEOs up at night: How to read the technology/digital landscape and spot potential disruptors, where to place big bets, and when to dive into the deep end. Are there lessons CEOs can take from DARPA’s approach to thinking about the future—and shaping it?

Dr. Tompkins: One key differentiator at DARPA is that we’re not afraid to fail. In fact, we embrace it because we know that if we always succeed, it means that we aren’t reaching far enough. That can be a tough position for a company to take, but tolerance for risk is important when you’re placing big bets. When we succeed, it can significantly change the world.

One note about changing the world: It’s not always about whether DARPA ultimately develops the technology. It’s about whether we give people the breakthrough they need to figure out what the next moves should be. With ARPANET (the precursor to the modern internet), DARPA program managers envisioned a world where humans and computers were working together. The ARPANET demonstrated the connectivity, the National Science Foundation then scaled it with NSFNET, and then the internet as we know it now really was born when policy makers enabled it to expand and freely evolve.

BLC: How do you think about talent and teams at DARPA, particularly in the context of diverse thinking and challenging assumptions? What kinds of diverse attributes does DARPA actively look for?

Dr. Tompkins: I’ve said more than once that the thing that keeps people at DARPA up at night—me most of all, perhaps—is the idea we don’t hear about or the person we don’t meet whose work holds the promise of a breakthrough technology for national security. DARPA is a collection of people. The technologies we so famously developed are the result of collective efforts—a lean government team, talented performers across academia and industry, and our partners in the military service labs. One of our greatest institutional risks is that the DARPA ecosystem could be too limited to government insiders, so we are actively investing in ways to expand that ecosystem, bringing in new performers as well as new DARPA program managers and support contractors from across the country, and growing our hiring pipeline upstream, so that in 5 to 10, maybe 15 years down the road, we will be able to draw from a broader and more diverse talent pool.

BLC: Does “thinking differently”—avoiding groupthink and other cognitive biases—come naturally to DARPA’s teams? Is there an ethos or mindset that helps to drive out-of-the-box thinking that’s required for breakthrough innovations?

Dr. Tompkins: DARPA tries to structure its programs around a question that naturally brings different communities and tech disciplines together—for instance, “can we build a prosthetic hand that would let you play the piano?” Of course, finding the kinds of people who can ask such questions is hard. As DARPA tends to do, and as I’ve mentioned above, we’re actively pursuing ways to find more of those people. As a recent example, last year we switched up DARPA’s approach to conference events. DARPA has regularly done an event every few years to get people thinking and to engage new people to work at, for, or with DARPA. In the past, these were single, really big events, typically held on the East or West Coast. A few years ago, I moved to Colorado, and in working with the R&D community there, I started to realize just how many people we were missing. Last year, we held a six-event series in research and development hubs across the country to try to close that connection gap and make it easier for innovators to spend some time getting to know DARPA.

We also launched several efforts to keep the momentum going. DARPAConnect is a new initiative with a goal to broaden DARPA’s reach and stimulate growth and collaboration with small businesses and education institutions that are new to the national security space. With a focus on underrepresented, diverse, and nontraditional performers, DARPAConnect aims to further break down barriers to entry. We are working with the Applied Research Institute to facilitate small events in cities around the country along with education opportunities, workshops, a mentor/ambassador component, and other outreach efforts.

One more example: Our new BRIDGES initiative is a pilot program to connect innovation from small companies that traditionally do not work with the United States government to classified Department of Defense R&D efforts. We lose out every day because the people who might have the best solutions are not even allowed to see the problems we’re trying to solve, and we think we can do a lot better.

BLC: Do you see a connection between the diversity of a project team and its agility/ability and willingness to change course—radically, if necessary?

Dr. Tompkins: I think that diversity allows a team to detect more quickly when change is needed, and to come up with a much broader range of options when they need to change course. Their willingness and ability to change course at all, in my opinion, is more tied to the culture and incentives of their organization.

BLC: Recognizing that DARPA (arguably) doesn’t get the same investor pressure for results that companies do, and your mission is ostensibly to “go for the breakthroughs/moonshots”—are there any lessons on risk-taking and R&D investment that companies can take from DARPA’s approach?

Dr. Tompkins: This overlaps heavily with our discussion of risk tolerance. One key reason we are able to tolerate risk is our rotation of people. Program managers are here for four to five years, racing against the clock to prove that the impossible is possible. No one stays long enough to be complacent or comfortable. The short tenure builds tolerance for risk.

Our willingness to celebrate failures along with successes is also an important part of the equation. When you’re not punished for failure, you’re much more inclined to dream up higher-risk, but potentially higher-payoff, ideas.

BLC: With AI tools like ChatGPT rapidly coming into mainstream use, how do you think about the upside and downside of using AI in innovation? How big of a concern should algorithmic bias be to technologists and innovators?

Dr. Tompkins: Large Language Models (LLMs) such as ChatGPT have significant potential for technological (and societal) surprise, so we are tracking them closely, and the behind-the-scenes state of the art appears to be significantly better than what is currently being shown publicly. So, while issues like bias are of concern, there is progress being made. In general, we’d encourage people to consider AI-based innovation as one component of a diverse team.

With respect to DARPA’s mission, national security needs don’t always align with industry needs, so we will continue to invest in national security-specific AI. We have multiple DARPA programs that are in the thick of exploring this technology as it relates to national security problems, and we take seriously—and we integrate into our work—the ethical, legal, and societal implications of the technologies we help create so that we are better able to provide policy makers with useful paths forward.

BLC: To what extent does DARPA look across industries—outside of tech—for signs of disruption and potential innovations/solutions on the horizon?

Dr. Tompkins: All the time. Business innovation is of particular interest as we strive to expand our innovation ecosystem and find ways to transition breakthrough technologies more quickly. So, give us your ideas! 

BLC: As someone who is focused on technology and the tech horizon for a living, how would you characterize corporate America’s positioning in the tech/digital space today? Where do you see U.S. companies falling short or perhaps being caught flat-footed over the next three to five years? Any thoughts on companies being reluctant to threaten or sacrifice their existing products and services in the pursuit of breakthroughs/blockbusters?

Dr. Tompkins: This is a question I might turn back to your readers. Historically, many companies had large R&D departments. A trend now is for that large company to buy and absorb a smaller company developing a desired innovation or technology. Sometimes they’re interested in only a single product line, but everything else that small company was working on disappears from the ecosystem. What impact does that have on competitiveness and innovation?  (By the way, I have seen technologies being bought specifically to take them out of competition with an existing product line, and I worry about that.)

BLC: “Doing good” and helping to address pressing societal issues—climate risk and energy transition, diversity, equity, inclusion, and the wealth gap—is increasingly being seen as good for business. From your experience collaborating with the private sector at DARPA, are you optimistic about public-private efforts yielding some breakthrough solutions for society any time soon?

Dr. Tompkins: I am definitely optimistic. I think the rare earth mining operation in Mountain Pass, California, is an interesting study on this topic. This isn’t work that DARPA is directly involved in, but certainly aligned with the supply chain and rare earth elements challenge DARPA is approaching from a few different angles.

The Department of Defense has invested in the site in support of U.S. supply chain resiliency. The approach relies somewhat on a belief that the nation will be willing to invest in environmentally safe production, as well as for the societal benefits of a resilient U.S. supply chain. The economics of mining are complex, so it is difficult to predict success, but I am cautiously optimistic that their approach, combined with technological breakthroughs that the government is investing in, could lead to good societal outcomes in the long run.

The views and opinions expressed herein are those of the interviewee and do not necessarily represent the views and opinions of KPMG LLP.