NASA Frontier Development Lab
What is AI? If you trust the ads you see on the train, in your browser, it's magic. Everyone is using it yet few can explain what it's really doing. What I'm certain of is that AI, is a way leveraging advanced computing hardware to pipeline mathematics and statistics to solve problems in society. At least, that's the definition that I care about the most. While I'm excited to see faster and faster AI algorithms, outperforming on benchmark datasets like handwritten numerals (MNIST), what I'm really excited about is how such computation can catalyze interdisciplinary science and solutions to social problems. We have to apply AI, not just make it faster for its own sake.
What happens when you make a bunch of scientists and AI experts live and work together in intense, dynamic teams for 8 weeks? I've gotten a rare chance to find out, being unexpectedly accepted to the NASA Frontier Development Lab program. Called a "science acceleratory", FDL pits computing and science and enginnering experts in close teams against some of the toughest challenges in space science. I will be working on the astrobiology team, serving for lack of a better term as a translator between the space science and AI sides of the table. I'm really humbled to be here, and hoping we can leverage computing power generously provided by our partners Google Cloud and NVIDIA, with high-risk-high-return brainstormings from our team of 1 astrobiology planetary scientists, 1 astronomy data scientist (me), 1 machine learning expert, and 1 big data particle physicist, against our challenge "What is universally possible for life?" We are going to try and extract, from earth and data of its most extreme life forms, what general patterns could be true for life elsewhere. How can biology stabilize a planet? How can it destablize? Over what timescales do planets coevolve with their biospheres?
I will be staying at the NASA Ames research center lodge, and working day-to-day at the SETI institute in Mountain View, CA-- after a 1-week AI*science bootcamp at the new NVIDIA "Endeavor" headquarters. Only 8 weeks (until August 17th) to find the answers to exo-life: plenty of time, right?
