As happens in many occupations, what used to require humans eventually becomes doable by machines. Add to that list of scientific skills the ability to revisit and refine previous research. Two years ago a robot was able to successfully form and test hypotheses. That robot, Adam, has now demonstrated the ability to refine its work when given new information. As WIRED Science reports, the robot was given new information on yeast growth rates relevant to previous work it performed. With this new information Adam could then do new experiments based on its work and determined that knocking out certain enzymes improved the rate of yeast growth.
While that’s a valuable conclusion (assuming it’s successfully replicated), the methodology behind this new work – from a computing perspective – is the potentially radical development.
What was critical to Adam being able to revisit its previous work was capturing and storing that work in a precise, standard language. If more computers and robots are to digest research findings, conduct and replicate their own experiments (or the experiments of others), they will need a standard language – one that works across the field, or across all fields.
This may be easier said than done. When I posted about Adam’s 2009 exploits, I figured at best you could get a robot specialized to a particular field or subfield. That reflects my concerns over the difficulty of getting everyone to agree on the same, exact meaning of words like trial, protocol, test, and experiment. This may seem easy, but each of those terms cover a multitude of different things. It may not be possible to make a research robot that could be as productive as a human researcher, but able to work in different, even unrelated fields. While researchers may breathe a sigh of relief at this point, these robots will likely be able to do trials and experiments much faster and more reliably than human could. So those tenure track jobs are likely to shrink further over the coming years. Maybe these robots could make up for the sheer numbers of (likely lesser quality) researchers other countries are producing.