Solving problems the market can’t

We caught up with Jason Bono, a former particle physicist who, as a consultant, has helped ARPA-H apply AI to improve healthcare, advised the Pentagon on strategies for emerging technologies, and built innovative ML systems used across numerous federal agencies to supercharge fraud and crime prevention. He is currently starting a new AI venture as part of a founding to give incubator. Jason shared that consulting improved his leadership skills and believes consultants working in quantitative domains would do better work if they defined their work as precisely as scientists do.

So, you started your career as a particle physicist. Why leave?

Five years ago, I was working as a research scientist in particle physics, and what spurred me to move on was a growing realization that the field was plateauing. There were plenty of job opportunities, and I could have made it my full career, but I felt like I wasn’t creating meaningful new knowledge on the fundamental nature of reality, which is the whole reason I pursued physics.  My sense was that many of the most interesting questions in 21st-century physics are proving too difficult to make real progress on within the timescale of a career. I hope I’m wrong and that a new renaissance comes soon, but I didn’t want to bet my career on it. 

I started thinking it was time to work on more pragmatic problems with direct societal impact on potentially shorter time horizons. That's actually how I found resources like Consultants for Impact in the first place. I was doing a rethink of what the big problems were.

It couldn’t have been easy to switch into consulting. How did you end up making the switch? 

I took an intermediate step first: working at the office of the Secretary of Defense through a tech policy fellowship. I received a security clearance and got to work on some pretty interesting problems related to emerging technology and strategy for the US military –  like doing consulting for the Pentagon.

I found advice from 80,000 Hours and Niel Bowerman really helpful when I was deciding whether or not to do it, and I’m glad I did. I had a belief that when government takes on a problem, it usually means market mechanisms have fallen short - which makes these issues particularly important to address. The market does incredible things, but I wanted to look for problems that the market isn't solving.

At Deloitte, I led teams to build machine learning applications for federal agencies, including the treasury, and in healthcare, defense, and intelligence. It was an awesome experience – not only did I get to solve what I thought were important problems, but I also learned a lot about business in general, which rounded out my scientific experience. 

I bet! You’re no longer at Deloitte, so what are you focusing on now? 

In July 2024, I transitioned into a role supporting the US Advanced Research Projects Agency for Health (ARPA-H). After surveying a range of government problems, I was eager to focus on moonshot efforts to improve healthcare using AI. If you're focused on improving healthcare, you can wholeheartedly throw yourself into creating technological advancements without worrying about misuse. 

I am currently  working on building a new venture as part of Ambitious Impact’s first founding to give cohort.  I will be donating the majority of any future earnings to effective charities through an agreement that’s similar to Founders Pledge’s.  

What did your time in consulting teach you? 

I was able to sharpen my ability to lead a diverse group of people. When you work with and lead a number of teams across fields, you start seeing patterns in what makes teams operate well – and not operate well! I was able to identify patterns on where dysfunction begins and ask, what does a good team look like? 

In the past, I worked with just physicists and scientists. I would have to interface with funders, lawyers, industry experts,  and other partners, but that's very different than collaborating daily with people from different disciplines.  I learned that if your thinking is clear but you're not communicating clearly, it's amazing how often it leads to chaos.  I've seen teams that are running in circles because everyone means something different by the same words, so I became fairly obsessed with precision language and defining new terms, especially in software development. To get everyone on the same page, I agonized over the details of the new lexicons we were creating, which paid off over and over.

Looking back at your time in consulting, what do you wish you had known on day one?

I don't think I had enough confidence! I was a lot more prepared for consulting challenges than I expected.

You're also working on AI safety projects. Where do you see opportunities in that field for people with consulting backgrounds?

Organizations need help handling uncertainty in decision-making. I've noticed that even technical organizations can be slow to use common techniques to estimate uncertainty, even in high-stakes circumstances. They often use an oversimplified or static model in a way that leads to an inaccurate understanding, which robs leaders of the chance to make a better decision.

Imagine being a tax agency that wants to make revenue projections. If you estimate average incomes and don’t include the richest man in the world, you’ll get a very different number. This type of thing happens at every scale: organizations making decisions based on averages without factoring in the volatile, extreme examples on the ends. Making good decisions related to AI is very hard to do when you exclude the tail-end scenarios. 

Using better techniques to quantify uncertainty isn't just a technical issue - it's about improving institutional decision-making. Consultants are well-positioned to help organizations develop better frameworks for handling uncertainty and making more evidence-based decisions.

Do you have any advice for consultants?

Be thoughtful about what challenges are worth taking on. The scope of work on your projects matters a lot. 

This is really for leaders of consulting teams: if you want to succeed, don’t stop until you’ve identified a scope of work that will be both effective for the client and motivating for your employees! If you use sloppy thinking when defining a problem and approach, it will come back to bite you every single time. 

If you just accept the problem diagnosis blindly, you are probably going to contribute to the problem. In the long run, you’ll make clients much happier by genuinely doing good work. 

Cheers, Jason. Thanks for the inspiring work!

Thank you!

‍