self-driving-lab #exploration #experiments #discovery #opportunities

<p> I am fascinated by the idea of self-driven laboratories (SDL), systems that integrate synthesis and analysis and automatically work towards predefined targets. However, I wonder if we are automating our own methodological biases. </p> <p> Probably, we've all heard about the scientific method: we formulate an hypothesis, plan experiments to validate or invalidate it, revise, and repeat. </p> <p> But, honestly, who has ever followed such as streamlined approach through their scientific career? There's so much that comes from serendipitous exploration and a posteriori justification. </p> <p> I am quite confident that we all have kept doing measurements before having an hypothesis we wanted to validate. Probably we all found "cool data" before we knew what it meant, what we wanted to prove, or if it was novel at all. </p> <p> I still remember the day I went to the lab and found a small creature spinning inside my optical tweezer. I was so excited I kept acquiring images, making movies. Back then I still was using Facebook and posted about it. Of course it was not a breakthrough, a biologist friend quickly explained it was just bacteria that grew overnight in my flowcell. </p> <p> And I wonder if the nature of an AI-driven laboratory will target our scientific method as we claim we follow it, or the true randomness of exploration. </p> <p> Can AI break free from just being average? </p> <p> Don't get me wrong, automation is powerful in itself. Running experiments overnight, decision-making faster than what a person could achieve enables fantastic new results. </p> <p> But the idea of "scientific superintelligence", albeit compelling for what it could entail, may end up derailing the efforts that are worth doing: defining open interfaces for hardware and software, lower the barrier for scientists doing software development, clear data sharing protocols to allow decentralization, systems engineering focused on extendibility. </p> <p> I still believe there's a wave of change coming, I just don't believe it'll be a discontinuity in human discovery, just progress in a context of ever evolving and growing complexity. </p>