Report from the OpenAI hackathon | OpenAI
On March 3rd, we hosted our first hackathon with 100 members of the artificial intelligence community.
Listen to article
On March 3rd, we hosted our firsthackathon(opens in a new window)with 100 members of the artificial intelligence community. We had over 500 RSVPs arrive within two days of announcing the event—if you didn’t make it this time, please RSVP again in the future!
Thank you toCirrascale(opens in a new window)for providing GPU machines during the hackathon.
Our applicants included high schoolers, industry practitioners, engineers for nonprofits (not just at OpenAI!), researchers at universities, and more, with interests spanning healthcare to AGI. We could only accommodate one hundred people this time so we tried to pick a balanced crowd with a wide range of backgrounds and levels of experience. In particular, we strove to achieve gender balance; many attendees told us that this kind of representation made a positive difference for their experience of the hackathon.
After the talks wrapped up, the hacking began. Over the course of an 8-hour code sprint participants authored dozens of AI projects on topics ranging from safety to healthcare. Some of our favorites:
* Jiale Xian, Clarence Leung, Kyle Zheng, Madeline Hawkins, and Stergios Hetelekides worked on an image classifier to identify purine-rich seafoods that gout patients should avoid.
* Arthur Juliani(opens in a new window)implemented PPO with curiosity-based intrinsic rewards and trained an agent to smash block towers:
* High schoolers Ethan Knight and Osher Lerner worked onNatural Q-Learning(opens in a new window)based on one of ourrequests for research(opens in a new window).
* Malhar Patel and Lee Redden worked onAeroEnv(opens in a new window), a Gym interface for a physical robot.
* Rodger Luo implementedQ-learning in Processing(opens in a new window), a programming tool for media artists.
* Andy Matuschak developed thescrying pen(opens in a new window), a fun and unique tool for machine-assisted sketching based onSketchRNNs(opens in a new window).
* James Giammona and Brad Neuberg generatedideas(opens in a new window)for how RL agents might avoid dangerous parts of the environment that they haven’t seen before.