Human versus machine: AI narrowly defeats human scholar in scientific skill test

PISCATAWAY, NJ — No invention signifies the ingenuity and intelligence of mankind quite like the computer. A miracle of the modern age, countless works of science fiction have predicted an inevitable showdown in the not so distant future: man versus machine. Now, according to researchers at Rutgers University, it looks like machines have already beaten humanity when it comes to at least one science topic.

Professor Vikas Nanda of Rutgers University has spent more than two decades meticulously studying the complex nature of proteins, the highly complex substances found in all living organisms. He has dedicated his professional life to contemplating and understanding the unique patterns of amino acids that make up proteins and determining whether they become hemoglobin, collagen, etc. Additionally, Professor Nanda is an expert in the mysterious step of self-assembly, in which certain proteins stick together to form even more complex substances.

So when the study authors set out to conduct an experiment pitting a human – someone with a deep and intuitive understanding of protein design and self-assembly – against the predictive abilities of a computer program. ‘IA, Professor Nanda made a perfect participant.

The study authors wanted to see who, or what, could do a better job of predicting which protein sequences would combine best – Professor Nanda and several other humans, or the computer. The published results indicate that the intellectual battle is near, but the AI ​​program has beaten the humans by a small margin.

What can scientists use protein self-assembly for?

Modern medicine is heavily invested in the self-assembly of proteins because many scientists believe that complete mastery of the process can lead to many revolutionary products for medical and industrial use, such as artificial human tissue for wounds or catalysts for new chemicals.

“Despite our vast expertise, AI performed as well or better on multiple datasets, showing the enormous potential of machine learning to overcome human biases,” says Nanda, a professor in the Department of Biochemistry and Molecular Biology. from Rutgers Robert Wood Johnson Medical. The school, in a university outing.

Proteins are made up of large amounts of amino acids, joined together end to end. These chains of amino acids fold to form three-dimensional molecules with complex shapes. The exact shape is important; the precise shape of each protein, along with the specific amino acids it contains, determines what it does. Some scientists, including Professor Nanda, regularly engage in an activity called “protein design”, which involves creating sequences that produce new proteins.

More recently, Professor Nanda and a team of researchers designed a synthetic protein that can quickly detect the dangerous nerve agent known as VX. This protein could lead to the development of new biosensors and treatments.

For reasons still unknown to modern science, proteins self-assemble with other proteins to form important superstructures in biology. Sometimes proteins appear to follow a design, such as when they self-assemble into a virus’s protective outer envelope (capsid). In other cases, however, the proteins will seemingly self-assemble in response to something going wrong, eventually forming deadly biological structures associated with diseases ranging from Alzheimer’s to sickle cell anemia.

“Understanding protein self-assembly is fundamental to making progress in many fields, including medicine and industry,” adds Professor Nanda.

How did the AI ​​program perform?

During the test, Professor Nanda and five other colleagues were given a list of proteins and had to predict which were likely to self-assemble. The computer program made the same predictions, then the researchers compared the human and machine responses.

The human participants made their predictions based on their previous experimental observations of the proteins, such as patterns of electrical charges and degree of water aversion. Humans ended up predicting that 11 proteins would self-assemble. The computer program, meanwhile, through an advanced machine learning system, chose nine proteins.

The human experts were right about six of the 11 proteins they chose. The computer program achieved a higher percentage of accuracy, with six of the nine selected proteins able to self-assemble.

The study authors explain that the human participants tended to “favor” certain amino acids over others, which led to incorrect predictions. The AI ​​program also correctly identified some proteins that were not “obvious choices” for self-assembly, opening the door for more research. Professor Nanda admits he was once a skeptic of machine learning for protein assembly investigations, but now he is much more open to the technique.

“We’re working to get a fundamental understanding of the chemical nature of the interactions that lead to self-assembly, so I was concerned that using these programs would miss out on important information,” he concludes. “But what I’m really starting to understand is that machine learning is just one tool among many, like any other.”

The study is published in the journal Natural chemistry.

Leave a Comment