A decade of research versus 48 hours of artificial intelligence - who will win? 🤯🤖
Machine translation from the original post. Not reviewed by a human translator. Historical claims may no longer be current.
A decade of research versus 48 hours of artificial intelligence - who will win? 🤯🤖
For 10 years, a team of researchers led by Prof. José Penadés from Imperial College London tried to understand how superbugs become resistant to antibiotics.
Then they gave the problem to a new artificial intelligence tool from Google (DeepMind, which focuses on accelerated scientific discovery using artificial intelligence).
They gave it a short description of the problem - and within just 48 hours, the AI reached the same conclusion they had been working on for an entire decade.
But it doesn't end there - the AI not only reproduced the discovery but proposed 4 more new hypotheses, one of which even the researchers hadn't thought of!
Prof. Penadés was so shocked, "How can this be?", that he thought Google had somehow gotten into his computer... but no. It was simply the enormous computing ability of AI.
🚀 "This is going to change science - unequivocally," the researcher said, adding that working with this tool felt to him like "finally playing in the Champions League"
Following the AI's suggestions, the team opened new research directions.
💡 Is the tool available, and to whom?
Tools like AlphaFold are available to certain researchers and deal mainly with deciphering protein structures (which are also very relevant to understanding bacteria).
There are open versions of AlphaFold that scientists can use, but most of DeepMind's advanced tools are still not accessible to the general public.
So does AI replace scientists? Or is it the best tool they have ever received?
🤔 In the future, will scientists be more like "strategists", while the AI runs the experiments and conclusions for them? Similar to the revolution developers are experiencing with AI-based development tools that keep gaining popularity
