Harvard's AI co-scientists join three cancer research teams
Three Cancer Grand Challenges teams will test whether AURORA can contribute useful hypotheses and experiments.
An AI-generated scientific idea still has to survive an experiment.
What happened
On 6 November, Cancer Research UK detailed the next stage of AURORA, an AI co-scientist platform developed by a Harvard-led team. Three Cancer Grand Challenges teams will evaluate it in work on cancer initiation, immune recognition and CAR T-cell therapies for children.
Why it matters
The system is designed to analyse evidence and propose hypotheses or experiments. Human researchers retain scientific decision-making and will test its suggestions. That creates a way to assess whether its ideas contribute to active research, beyond generating convincing text.
The bigger picture
The evaluation is itself an experiment. Useful recommendations could help researchers decide what to investigate next, but the announcement does not demonstrate faster discovery or better patient outcomes. Those are questions for the research teams to answer through evidence.