AI chatbots solve mathematical problems beyond human capacity

A recent article published by The Guardian announces “the first genuine scientific discovery made by large language models (LLMS)”. It explains that DeepMind has introduced “FunSearch,” a system utilising LLMs to write computer programs that solve problems. Paired with an “evaluator,” FunSearch ranks programs by performance, combining the best ones to improve the LLM continually. The AI mo tackled mathematical challenges, producing solutions that surpassed existing human-derived ones. The collaboration of human mathematicians with AI, as seen in this research, opens avenues for innovative problem-solving and algorithmic advancements in computer science.

In what other areas do you think AI can exceed human knowledge? Let us know in the comments!

programming visual
programming visual

Find the full article from the Guardian here.