O Google just launched o AlphaEvolve, an encoding agent that uses the Gemini and evolutionary strategies to create algorithms for scientific and computational challenges — driving efficiency within the Google and solving historical mathematical problems.
ADVERTISING
AlphaEvolve Details
- AlphaEvolve uses a combination of models Gemini (Flash for idea generation, Pro for analysis) to create code, which is tested by evaluators and iteratively evolved.
- The system has already made several mathematical breakthroughs, including finding the first improvement to Strassen's algorithm since 1969.
- It is also increasing the efficiency of the Google, optimizing data center scheduling, improving AI training (including its own), and aiding in chip design.
- When tested on over 50 open mathematical problems, it matched state-of-the-art solutions in 75% of cases and discovered entirely new and improved solutions in another 20%.
Why does it matter
Mathematics plays a role in almost every aspect of life, and the pattern and algorithm strengths of artificial intelligence (AI) appear poised to unlock a whole new world of scientific discovery.
Read also
