New Google AI Chatbot Tackles Complex Math and Science

image via New Scientist

May 15, 2025, 2:30 PM

• •Google DeepMind's AlphaEvolve, combining LLMs and algorithms, tackles complex math and science problems.
• •AlphaEvolve improves AI chip designs, discovers faster matrix multiplication methods, and optimizes computing capacity.
• •The system uses an evolutionary approach, refining solutions through LLMs and evaluation algorithms, outperforming specialized AI tools.
• •While promising, wider accessibility and external testing are awaited, though DeepMind aims to share AlphaEvolve with the scientific community.

Google DeepMind's AlphaEvolve, a new AI system, is making significant strides in solving complex math and science problems. This innovative system combines the capabilities of large language models (LLMs) with algorithms to refine and improve solutions. AlphaEvolve has demonstrated its potential by enhancing AI chip designs, discovering faster methods for matrix multiplication, and optimizing Google's computing resources. Although promising, the system's wider availability and external validation are awaited.

Read original article