Scientists in the UK have incorporated artificial intelligence (AI) and a sophisticated set of algorithms into a robot, named Adam, that can develop hypotheses and carry out a whole scientific experimentation cycle. The research, published in the journal Science, paves the way for dramatically increasing the rate of scientific progress.
Currently, automation is used in laboratories to carry out high-throughput experiments, which has helped to process vast amounts of scientific data in a much shorter timeframe than would otherwise be possible. However, in the area of systems biology, which is complex and involves staggering amounts of data that could not reasonably be analysed by any single human being, such technologies are inadequate. For example, one person would have great difficulty analysing the volume of information in one genome, and would find it impossible to analyse several of them together.
According to the current study, led by Professor Ross King at Aberystwyth University in the UK, robot scientists have the potential to provide more than simple automation to systems biology. By originating hypotheses, devising experiments, physically carrying them out and continuously interpreting the results, they could revolutionise the way scientific research is carried out. Advances in AI and computational systems are making this possible.
The scientists developed a robotic system and set it the task of identifying the genes that encode enzymes which catalyse reactions in baker's yeast (scientists use this organism to model more complex life systems). This 50-year-old puzzle had not yet been solved, and the researchers were interested to see how the robot would tackle it.
Adam came up with and tested 20 hypotheses, and produced conclusions based on its experiments. The researchers then confirmed the results by repeating the experiments manually. The reason the problem had been so difficult to solve, they found, was that there were so many complicating factors - a common problem in systems biology. According to the study, the robot's analyses were necessary to the feat of unravelling this 'web of functionality'.
Describing and reporting science clearly is essential for the free exchange of scientific knowledge. Importantly, the robot was able to record every step of the process, from beginning to end, in great detail. Adam was also able to go through the entire scientific experimentation cycle in a very short time, which gives the researchers hope that their new knowledge will enable new discoveries to be made at a much quicker pace than is currently possible.
"If science was more efficient, it would be better placed to help solve society's problems," said Professor King. "One way to make science more efficient is through automation. Automation was the driving force behind much of the 19th and 20th century progress, and this is likely to continue."
While it could be argued that Adam was able to make its discovery because of the way the researchers had formulated the problem, the team believes that Adam is the first machine to have independently discovered new scientific knowledge.
"We accept that the knowledge automatically generated by Adam is of a modest kind," the study concludes. "However, this knowledge is not trivial, and in the case of the genes encoding 2A2OA, it sheds light on, and perhaps solves, a 50-year-old puzzle." Adam demonstrated how a simple form of hypothesis-led discovery can be automated - a significant step forward for AI and computational systems.
Using robot scientists in the laboratory will hopefully enhance the way human scientists study biology. Commenting on the study in a podcast interview with Science, Dr David Waltz said, "Clearly, in biology there are vast amounts of data that have to be understood. There is no good way to simplify biology down to a few simple terms. In some sense, biology requires that we catalogue, understand and organise vast amounts of data. Biology is uniquely in need of methods like the ones described in order to make some sense of it." The new findings also have implications, he said, for enhancing the study of astronomical data and climate modelling.
In the accompanying editorial, Dr Waltz and Bruce G. Buchanan commented, "Human-machine partnering systems that match the tasks to what each partner does best can potentially increase the rate of scientific progress dramatically, in the process revolutionising the practice of science and changing what scientists need to know."
The next step is to extend the robot's functions and capabilities by using software that will enable external users to propose hypotheses and experiments. The scientists hope to develop a way of enabling teams of human and robot scientists to work together. Adam is a still a prototype, but the UK team believes that their next robot will hold great promise for scientists searching for new drugs to combat diseases such as malaria.
For further information, please visit: