From Scientific American:
Scientists report that they have developed a robot that can formulate hypotheses, design experiments to test them and analyze the results. What is more, it performs just as well as real grad students and spends less money.
OK, so it’s not as bad as the entry title makes it sound. The real punchline here is not the bit about equalling grad student performance, which a monkey could do if my own grad school work is any indication; nor do I think that automatic methods of generating hypotheses and/or interpreting data have any real future, except in cases so simple that doing it with wetware is trivial anyway. The real kick is the potential for drastic reductions in monkeywork: one of these doohickeys could slave away 24/7 on the boring, repetitive tasks that make up the bulk of labwork. I hate that shit, and can’t wait to dump it on someone (or somebot) else. Another thing about robots: they don’t require initial training, and upgrades are presumably simpler than is the case with their vertebrate competitors; furthermore, a robot will not care if I am disagreeably sharp of temper and tongue1. Given how hard it can be to get good hominid help (just ask my former employers!), I can see a real market for these things.
1 I think of this as the uncharitable interpretation of my reluctance to suffer fools, gladly or otherwise; but then I would, wouldn’t I?