I graduated from Tufts University in 2016 with a bachelor’s in cognitive science and philosophy. In 2017-2018, I earned a graduate diploma and master’s in economics from the University of Cambridge. Currently, I’m a third-year PhD candidate at the Wharton School of Business. My goal is to improve the rigor and applicability of behavioral science using machine learning.
Machine learning and methods. My research on machine learning methods aims to solve a fundamental problem: Behavioral scientists favor simple linear models which are easy to interpret but inaccurate in predicting real-world behavior. Complex machine learning models accurately predict behavior but are difficult to interpret. To address this problem, I created Generalized Shapley Additive Explanations (G-SHAP); a technique in explainable artificial intellgience. G-SHAP allows behavioral scientists to gain theoretical insight from complex models which accurately predict behavior.
Cyborg forecasting. There are two ways to predict the future: expert human judgment and statistical/machine learning models. Human judgment can answer complex questions, but is mired by heuristics and biases. Models are highly accurate, but require well-defined questions and large datasets. My long-term research goal is to create human-computer ‘cyborg’ forecasting systems which combine the best in human expertise with the best in machine learning to predict critical geopolitical events.
In 2020, I released an alpha version of hemlock, a python package which aspires to be the most powerful and flexible way to create studies, with applications in marketing and behavioral science research.
I had the idea for cyborg forecasting systems in the first year of my PhD. But even if I built a machine learning tool to interface with human experts, I realized I would have no way to test it in a behavioral study. To solve this problem, I spent every evening and weekend during the first two years of my PhD building a powerful survey creation software. One of hemlock’s defining features is its ability to run interactive machine learning algorithms in real time during behavioral studies.
Effective altruism. Find out how to do the most good, then do it. I believe ensuring humanity has a future is the most significant challenge of our time. I plan to use cyborg forecasting systems to predict and preempt existential catastrophes.
Solving problems is more important than understanding them. Theory is intrinsically interesting and instrumentally important. Fundamentally, what matters is making the world safer and better. Theory is important insofar as it contributes to solving problems. But theory isn’t always necessary. It’s possible to understand that a solution works before understanding why it works.
Full stack academic. Papers are not the end product. Solutions are. I believe the gap between academic discovery and implementation is too large. My aim is to either partner with practitioners to make sure my work is used, or level up to become a full stack academic: taking a concept from hypothesis to deliverable.
A good tool makes simple things easy and complex things possible. Simple is better than complex, but complex is increasingly necessary. To facilitate work at every level, we need tools that allow us to execute simple tasks easily while granting us the power and flexibility to pursue complex ideas. This is the philosophy powering hemlock.
Technical sophistocation is necessary for scientific progress. Every scientific field has become and continues to become more technically sophisticated. Behavioral science is no exception. The low-hanging fruit has been picked - new biases are no longer waiting to be discovered around every corner - and technically sophisticated methods are necessary for the field to progress.
The only thing that matters is what you create. Your accolades and alma mater aren’t going to change the world. The only thing that matters is what you create, whether it’s an invention, a software package, a paper, or a work of art. You can even create an inspiring idea or a lifelong friendship. Forget the scholarships, the prestige, and the medals. Those will come. Learn the skills you need to create something extraordinary, then make it.
Have a vision. Strive for direction and coherence in your work. Everything you do should aim at a single high-level goal. Dedicate time to constructing, clarifying, and reevaluating that goal. Learn the skills and acquire the tools you need to pursue your vision - even if you have to invent the tools yourself. Once you have your vision, allow yourself to modify it, but don’t allow yourself to give up on it.