In an experiment that pitted an Artificial Intelligence (AI) program against researcher Vikas Nanda, the AI won with a close call, demonstrating the potential of machine learning. The study revealed that AI performed as well as the researchers and even better on several occasions.

Dr Vikas Nanda, a researcher at the Center for Advanced Biotechnology and Medicine (CABM) in the U.S. has studied the format of proteins for more than 20 years now and his expertise in the self-assembly of proteins prepared him as the perfect participant for the experiment.

Also Read | SpaceX launches Falcon Heavy on US Space Force mission after 3-year hiatus

For the research, Nanda and five other scientists from around the United States were presented with a list of proteins and were asked to predict which ones were likely to self-assemble. The AI computer program was given the same list to make its predictions as well. The predictions were then compared with each other.

The research team aimed at having humans compete against a machine, specifically an artificially intelligent computer program. The goal was to test instinctive knowledge and comprehensive information about protein design and sequences, against the AI’s predictive capabilities based on datasets and programming. The researchers made their predictions based on observation of protein behavior such as patterns of electrical charges while the machine based its knowledge on an advanced machine-learning system.

Also Read | What are death cap mushrooms?

The researchers predicted that 11 proteins would self-assemble in contrast to the AI computer program, which predicted that nine proteins would self-assemble. Both man and machine were correct for six of their respective selected proteins.  But because the computer program had a higher prediction percentage, it was able to beat the researchers by a close margin.

“Despite our extensive expertise, the AI did as good or better on several data sets, showing the tremendous potential of machine learning to overcome human bias,” said Vikas Nanda.

Also Read | At $8 per blue tick, is Elon Musk’s digital town square about to turn into a billboard?

Nanda continued, “We’re working to get a fundamental understanding of the chemical nature of interactions that lead to self-assembly, so I worried that using these programs would prevent important insights. But what I’m beginning to really understand is that machine learning is just another tool, like any other.”

The study has made Nanda, once a doubter of machine learning, more open to new-age techniques. The study was published in the monthly scientific journal Nature Chemistry.