AI SCIENCE
The lab assistant never sleeps because, well, software
AI is starting to do more than summarise research. New “AI co-scientist” tools can search papers, compare ideas, suggest hypotheses and help design experiments.
Google’s Co-Scientist was recently given a difficult cancer problem involving MYC, a protein linked to many cancers.
After researchers corrected a few early mistakes, the system searched more than 700 papers and came up with 108 possible approaches before narrowing them down to one unusual idea.
Here’s what you should know:
AI can explore several research ideas at once
It could speed up the early stages of scientific discovery
Human judgement is still needed to check what actually works
AI speedran the literature review
These tools are already being tested across biology and medicine.
In one case, Google’s system reached a conclusion about how bacteria exchange DNA in around two days, after researchers had spent years studying the same problem.
But humans are still very much needed.
AI can misunderstand data, suggest weak experiments or come up with ideas that are too expensive or impractical to test.
Scientists still need to ask the right questions, spot mistakes and test the results in the real world.
AI generating an incredible scientific breakthrough and then finding out the PhD budget is £14 and a Tesco meal deal.- MG