The discovery of 12 new functions for genes in one of the most studied organisms in the world wouldn't be news, except that scientists didn't discover them. A robot named Adam designed, carried out and discovered the new gene functions.
"Our goal is to make science more efficient," said Ross King, a professor of biology and computer science at the University of Wales and author of a new paper in this week's issue of Science detailing Adam's work.
"If we had computers designing and carrying out experiments we could get through many more experiments than we currently can," said King, adding "robots don't need to take holidays."
The 10-year-old Adam, which is housed at Aberystwyth University in the U.K., might replace humans eventually, but it doesn't look like one. From the outside Adam is 45 cubic meters of elongated white plastic instruments.
Inside Adam sits a biological library of more than 12,000 chilled petri dishes. Each dish contains a different yeast strain with various genes removed from them. With its various mechanical tools, Adam can grab the petri dishes, remove a sample of yeast, grow it, clean it and analyze the results of the experiment.
Adam actually discovered more than 12 new gene functions. When King and his colleagues compared the functions of all the genes Adam found, they realized that some of them had previously been described. So Adam had independently confirmed those results.
Adam is still a prototype, but King's team hopes their next robot, Eve, will help boost the search for new drugs to combat diseases such as malaria.
"This system is still a prototype," explained King. "The first car wasn't as efficient as a horse."
Adam and Eve not only have the hardware to physically manipulate objects, they also have advanced artificial intelligence systems that let them make their own decisions and then act on those decisions, without help from their human creators.
In another article in the current issue of Science, scientists from Cornell University trained a computer to watch the natural world and to come up with its own natural laws, or instances where something will always happen. Specifically, the scientists programmed the computer, which had no prior knowledge of physics, to independently produce the laws of energy and motion using a simple pendulum.
Both of the artificial intelligence systems are still relatively slow to learn, however. The Cornell computer took 30 to 40 hours of powerful processing to derive the laws of motion. Adam took three months to discover the 12 new gene functions, although it was working on other projects as well during the same time.
Nonetheless, the advanced artificial intelligence systems developed by U.K. and U.S. researchers, while still on the slow side, will increase the amount of scientific research that can be done, says David Waltz, a professor at Columbia University who was not involved with the Science study.
"We have vastly more data that we can gather and vastly more powerful machines that we can work on it with and we must create more storage to save them," said Waltz.
"We are able to now build systems that can explore scientific hypothesis largely independently, rather than have people examine it by eye or by mathematical methods."
All this means is that the jobs of graduate students and post-docs across the world are safe, assures King, at least for the next 10-20 years.