A new AI research lab aims to solve the industry’s most pressing challenges, inspired by the animal world, such as the movements of a swarm of fish or the coordination of a bee colony. While many leading companies in the field are trying to outdo each other by building better than ever artificial intelligence systems,

Sakana, whose name means fish in Japanese, believes it can do more with less data. The startup plans to build several small AI models, the technology that powers products like ChatGPT, and make them work together. The idea is that a “swarm” of programs can be just as smart as the large-scale efforts of larger organizations.

Founded by two of the field’s leading researchers, former Googlers David Ha and Leon Jones, Sakana’s approach could potentially lead to artificial intelligence that is cheaper to learn and use than current technology. That includes the prolific artificial intelligence that has taken Silicon Valley by storm with its ability to spit out text and images in response to prompts. The new startup’s approach is different from that of companies like OpenAI, which can feed all its data into a single large AI program rather than a series of small programs.

“Ants move around and dynamically build a bridge on their own, which may not be the strongest bridge, but they can do it right away and adapt to the environment,” Ha said. “I think this kind of adaptation is one of the very powerful concepts we see in natural algorithms.”

Ha and Jones are well-known names in the AI ​​research world. Jones, a Tokyo-based AI researcher, is the co-author of Attention All You Need, one of Google’s most influential articles in the field and the foundation for many of today’s most popular AI products. Ha, who also lives in Tokyo, was previously head of the research department at Stability AI. Prior to that, he focused on productive artificial intelligence as a scientist at Alphabet Inc.’s Google Brain. in japan

Sakana is still in its early stages: it has not yet built an AI model and does not have an office. It is scheduled to open in Tokyo soon, according to Ha. The company declined to comment on the fundraising status.

But the ideas that Sakana works with are more established. Near the end of his time at Google, Ha and a colleague started a project called “sensory neuron as transformer” and deployed a group of small AI models to work together in games, rather than using a single large model. Other researchers were also inspired by the workings of the human brain. For example, the term “artificial neural networks” refers to artificial intelligence models that are programmed to process information by trial and error, roughly in the same way that humans do.

“The human brain still outperforms our best AI,” said Jones. “So it’s clear that the human brain is doing something right that we don’t fully understand yet.”

Jones and Ha sat side by side at Google’s Tokyo offices, and the two remained in touch after Ha left the company. After working for an Internet giant for years, they finally turned to startups. Ha’s role in Stability AI meant he spent a lot of time building research teams, and he was eager to get back to research. And Jones felt limited at Google.

“Unfortunately, you could say I have a lot more speed outside of Google,” he said, noting that the need for permissions and resources can slow down the process of working on innovative technologies in a large company. Oh, when he suggested they start a startup, he said, “It made sense to me.” Source