Howie Altman (‘07-‘14) and Milton Hernandez (‘08-‘14) met in CS Tech and built a rapport that's extended far beyond Bridgewater's walls. Now, they've partnered as the brains behind Perceiver AI. We asked Howie a few questions about his time at Bridgewater, the problems Perceiver is taking aim at, and how they're opening new industries to AI.
*Edited for brevity and clarity
Q: How did you two come to work together on Perceiver?
A: We met when I joined Bridgewater's CS Tech team. Milton was already on the team as a senior software engineer. Over the years while we worked together at Bridgewater, we became close. After we both left, we continued to work together at different companies and on different projects. When I left Hello Alfred in early 2018 to start our first company together, Actualize Partners (a strategic consulting firm focused mostly on NYC startups), we decided to also roll in the Perceiver product and form Perceiver AI as the company through which we would offer it as SaaS.
Q: When you started working on Perceiver AI's technology, what problem were you trying to solve?
A: Virtually all existing AI technologies and applications are based on neural networks. It is our thesis that to, achieve general artificial intelligence (where a machine can perform all the same tasks as a human) and then artificial consciousness, other approaches are also needed. One very promising approach was genetic programming, or essentially the digital equivalent to biological evolution and natural selection. However, until Perceiver, there were a set of unsolved fundamental problems, which kept it from realizing its potential. We have solved these and Perceiver is the first implementation of genetic programming to be usable on complex, real world problems.
Another critical problem is that a neural net starts from scratch every single time. That's not how we evolved. If every generation started from scratch, we wouldn't get very far as a species. So how does Perceiver change this? You start with all the societal knowledge you can throw at it. Then, you build more over time as it learns, evolves, and factors out new salient elements to go into the body of societal knowledge. Instead of a relatively low, finite limit on how much a neural net can "learn," you have no limit. Learning accelerates over time, exponentially. Eventually, you have hundreds of trillions of generations of learning evolution powering decisions and predictions, radically compressed in time by our highly scalable distributed GPU computing architecture.
Q: What Perceiver pilots in new industries are you most excited about?
A: There really are too many to list. Personalized medicine is a huge emerging industry and AI will power a formidable part of it. We are scheduled to run a pilot in partnership with the Memorial Sloan Kettering Cancer Center, focused on personalized cancer treatments. "Optimizing the world" is our mission at Perceiver and we are currently working with Blue Prism, one of the leaders in Robotic Process Automation, as well as Airops, one of the leaders in the business aviation software space, to solve complex optimization problems like reducing infrastructure costs and job wait times, and minimizing fuel spend on flights. And we're scheduled to present to the team at the Princeton Plasma Physics Laboratory to identify an initial use case in practical fusion energy R&D.
On a related note, we are in the middle of raising our seed round of funding to begin ramping up our pace and reach. We have the honor of some members of the Bridgewater extended family as investors and would welcome more participation. We'd rather close the round with as much capital from F&F and angel investors who have a passion for what we're doing than anything else.
Q: Are there ways that your leadership approach at Perceiver is influenced by your time at Bridgewater?
A: My experience at Bridgewater was one of the most rewarding periods in my career. I had the good fortune of being part of the senior management team that initially worked with Ray on v1 of Principles and also had a front row seat to how it evolved as the organization evolved and as we worked through various growing pains. Many of the important Bridgewater values and principles are reflected in our way of being with one another. It's the expected standard throughout our organizations.
But more importantly, I know I learned more about myself, including identifying and dramatically improving on my biggest weaknesses, than I would have anywhere else – and in a lot less time than it would have taken anywhere else. I got to work with and be surrounded by some of the best and smartest people in the world, many of whom I would have never had the chance to meet anywhere else.
Q: Are there any other skills you honed at Bridgewater setting you up for success now?
A: Learning so deeply about human psychology and how it plays into getting what you need out of people with very different ways of thinking and perspectives has been a huge advantage. I am grateful to have been a part of Bridgewater's history and hope that I was able to leave things in a better place than when I got there.