Google DeepMind has introduced AlphaEvolve, a new AI coding agent that enhances large language models (LLMs) like Gemini to solve complex computing and mathematical problems. Utilizing evolutionary cycles, AlphaEvolve proposes coding solutions, tests them, and iterates to develop improved versions. This self-evolving process has already led to advancements in energy efficiency at Googleâs data centers and breakthrough solutions in mathematicsâ notably enhancing the known solutions to the centuries-old kissing number problem. AlphaEvolve marks a significant step in AI's application for complex problem-solving.
AlphaEvolve is a self-evolving AI that improves code solutions through evolutionary cycles, greatly enhancing Google's ability to tackle complex computing and mathematical problems.
In 20% of cases tested, AlphaEvolve improved on the best-known solutions to over 50 open maths problems, including a breakthrough in the 300-year-old kissing number problem.
Collection
[
|
...
]