Sakana AI is engaged in the search for the IA that is improving itself

The startup Sakana AI has put in place a new initiative unique in the world of AI: a laboratory dedicated to the research of the I.A. able to improve oneself via a technique called RSI (Recursive Self-Improvement). This approach contrasts with the large American labs that have a "race for compute", which reveals the growing need for more efficient and adaptable AI technology.

The concept of the I.A. which is improving itself has already been explored in previous works, notably by Llion Jones, co-author of the Transformer. This Sakana AI initiative is a response to the need to find safer and more effective alternatives than the current "race for compute".

For Anthropic, who criticizes this technology, the importance extends beyond its control potential: it also raises questions about the security and legitimacy of developing an AI capable of spreading. The position of Sakana AI, which sees the benefits of an ever-changing AI, is a challenge for an ever-changing industry.

This Sakana AI initiative raises crucial questions about the future direction of AI and its potential impact on the world of technology. His involvement in this research suggests a new perspective on the development of an AI capable of improving itself, contrasting with the great American labs currently engaged in a "race for compute".

---

Source: The Decoder. IA-assisted editorial summary — TechnoExpress.