OpenAI, a non-profit research lab, has released dozens of papers since its founding two years ago. However, one article posted online this week stood out: the first author is still in high school. That student is Kevin Frans, a high schooler currently applying to colleges. At just 15, he built his first neural network—a system used by tech giants for tasks like voice and face recognition. Inspired by stories about AI in games like Atari and AlphaGo, he started reading and experimenting with academic papers. "I enjoy making computers do things that were once thought impossible," he says. One of his projects is an interactive web page that generates comic-style line drawings.
A photo shows Frans working on AI research at OpenAI. He joined the organization after coming across their list of open questions. He shared some insights and even offered suggestions to researcher John Schulman, who was stuck on a problem related to trust policy optimization. Schulman was surprised to receive such thoughtful input from a high school student. Later, Frans met Schulman during an interview at OpenAI. This summer, while interning in San Francisco’s Mission District, he was the only intern without a graduate degree. He's now tackling a tough challenge in AI: how can machines solve new problems using what they've already learned?
Humans do this effortlessly. When cooking for the first time, we don’t relearn how to mix ingredients. But machines usually need to retrain from scratch for every new task. Frans, along with Schulman and researchers from UC Berkeley, recently published a paper addressing this issue. He believes solving it could be a major breakthrough for robotics and AI. He developed an algorithm that helps virtual robots learn movements applicable to multiple tasks, like walking or crawling. In tests, the algorithm helped bipedal and quadruped robots adapt faster. A video of an ant-like robot was submitted to ICLR, a top machine learning conference.
Beyond AI, Frans is also into martial arts, holding a black belt in Thai boxing. His passion for AI likely stems from growing up in Palo Alto, Silicon Valley. While his parents didn’t help with his projects, his father worked in semiconductor design at Xilinx. Olga Russakovsky, a Princeton professor in computer vision, called Frans’ achievements rare. She noted that AI is more challenging than traditional subjects because of the computational demands. When his desktop wasn’t powerful enough, he used his debit card to access Google Cloud. “The best way is to try it yourself,†she advised.
Researchers are encouraging young minds like Frans to contribute to AI. Russakovsky co-founded AI4ALL, a program supporting diverse high school students in AI. She believes AI must reflect society as a whole. Meanwhile, Frans is thinking about helping the next generation—like his 7-year-old brother, who he thinks might be interested in programming. “Maybe when he grows up, I can help him,†he said.
Flat Wire High Current Inductor
Flat Wire High Current Inductor,Small Flat Wire Common Mode Inductors,High Current Flat Wire Inductors,High Power Flat Wire Inductors
Shenzhen Sichuangge Magneto-electric Co. , Ltd , https://www.scginductor.com