ML Researcher, Janelia
If you think you can't do something, really question what you believe that.
WHAT DO YOU DO?
I research intelligence-- creating algorithms that mimic (or surpass) existing intelligence. How do you get computers to mimic-- and surpass-- what humans do?
Currently, I work as a researcher in theoretical neuroscience and machine learning, creating tools to make decoding the brain more efficient. Previously, I was a natural language processing (NLP) research engineer, devising better search engines than what we have currently. My experiences all center around researching intelligence broadly. I’m interested in fundamental questions: after the deep learning paradigm in computational neuroscience, what comes next? How far can we push the AI and brain metaphor? Can we create intelligence that surpasses human intelligence, with a greater representational capacity? How accurately can we model-- and change-- our reality? My questions are rooted in science and metaphysics. Answering them is not a job, but driven by intense curiosity.
WHY DID YOU CHOOSE THIS FIELD?
In childhood, I was ravenously curious about the fundamental mysteries of outer space and the mind. I was hungry to understand the nature of our reality-- whether it be extremely far away or deep within ourselves. I was also lucky enough to have a paradigm-shifting father, a civil-engineer by training, who was not afraid to ask bold questions that society starts discouraging you from asking: what is the nature of our reality, what is generating the universe, can we create tools to help us ask even deeper questions?
After getting mired in molecular neuroscience in undergrad at Princeton, I realized that artificial intelligence instead could clarify these questions. I entered the field at the height of the machine learning boom, with the hope that the representational capacity of AI could exceed that of humans. It became clear that neuroscience is but a subset of intelligence-- the brain is one instantiation in a vast possibility space of intelligences that could exist. Thus I moved from studying solely the brain to more general principles of information processing.
There are curious philosophical and cultural differences between these two frames: to a large degree, neuroscience is the messy project of reverse-engineering the brain, while computer science creates intelligence from first principles. To generalize, neuroscientists care about biological details-- which specific system is active, what is the concrete mechanism? There can be a lack of generality. In contrast, computer scientists will forgo details for general principles, but at their worst, get lost in pet theories and lose track of real intelligence (like the GOFAI scientists from a generation ago). Thus I camped myself between both fields, and took inspiration from the brain while not constraining myself to biological detail.
I was further intrigued by entrepreneurial culture, and spent some time as a machine learning engineer in the Bay Area startup world. The synthesis between action-biased startup land and the high intellectual standard of academia is formidable; I take inspiration from both. Private research companies like OpenAI and DeepMind, which deliberately cherry-pick the best of both qualities, are inspiring to me. Thus I maintain circles in both worlds and consider starting a research company in the future.
WHAT DO YOU LOOK AT & THINK, "I WISH YOUNGER ME WOULD HAVE KNOWN THIS WAS POSSIBLE?"
My achievements do not sit cleanly in one field-- I would have told my younger self to not restrict herself, not to feel any pressure "to be" anything too specific. While we hear about "world class literary authors" and "brilliant scientists," our current society discourages the Renaissance person, implicitly telling us that we cannot have it all. My younger self would have been inspired by the breadth of achievements: in memoir-writing, fiction-writing, neuroscience, computer science, machine learning, and finance. If you follow your interests and your passions, you can become something that society thought was impossible.
WHY DO YOU LOVE WORKING IN STEM?
I love the groundedness of STEM, and the consensus that there is an objective reality and that we can work together to understand it. We (mostly) agree upon scientific principles like falsifiability-- how often does such agreement happen in other areas of our society?
BEST ADVICE FOR NEXT GENERATION?
My greatest enemy has consistently been myself-- at my worst, I catch myself thinking limiting beliefs, disqualifying myself from opportunities before even pursuing them. I recommend listening to motivational videos on Youtube or Spotify as soon as you wake up-- the patterns of success then become apparent. Champions are relentless, disciplined, and most crucially, they believe in themselves even if no one in the world does. If you think you can't do something, really question what you believe that. It's probably a limiting belief that is holding you back.
I'd also tell girls to not listen to societally proscribed narratives about what you "should" be. This includes media depictions of "the unconventional, rebellious women," which is also a societal narrative! Instead, if you follow your drives and values, and study what works across different environments, you can become something novel that society thought was impossible.
I would also warn that socialization can be dangerous as a girl (Reviving Ophelia is a great book documenting this)-- and to be aware that very subtle microactions, like wearing a dress, can create strange dynamics in, say, a room of male physicists. This is not to say not to wear a dress, obviously. Rather, trust your observations, and always follow your intuition-- it is usually correct. If a situation seems weird due to gender dynamics, it probably is (society can gaslight you into thinking that your feelings are false). At the same time, do not be too paranoid-- there are also many kind men out there who will want to help you. Find real women role models, as most media depictions of women are corrupt (although this is getting dramatically better).
INSPO / FUN FACT
"Possunt quia posse videntur" (They can because they think they can) - The Aeneid