In laboratories all over the world, AI tools help researchers in sorting, reacting, and even creating those that look like understandable assumptions. First, he was auxiliary, then a partner. Now the lines are rapidly fading. The machines are learning to ask their questions, and in some cases, design their own experiences.
A recent Nature The article re -attributes the tension, which raises a bold question: Can AI ever win a Nobel Prize? Have we finally arrived at the stage where this prestigious award goes to a machine or software? Well, now it does not seem as unacceptable as it may happen a few years ago.
From the prediction of the chemical reaction to the biological paths, the AI models are approaching sovereign discovery. Some researchers say it is just a matter of time unless a machine works for Nobel. The only thing is less clear whether we are ready to give him the Nobel Prize for his work.
The most important step associated with this question is the Nobel Touring Challenge. Launched in 2016 by Sony AI’s CEO Heroki Ketano, which sets a promising purpose for the future of AI in science. This raises the same question we are now asking: Can we create an AI that can be given a Nobel Prize for our discovery? Not as a helper, not even as a co -author, but as the main architect of the entire scientific process.
Successful SUCH, such a system will have to identify its research question, design your experiences and have your own experiences, realize the consequences and realize something new. It will need to be fully independent of new ideas or routes that even human researchers did not expect nor understood. So far, this has not happened.
However, according to some we know the moment we know. “Human level painting is a challenge,” says Ross King, one of the AI researchers and organizers of the University of Cambridge. He believes that the same is the case for scientific achievements. It may take decades, or it may arrive very quickly.
AI’s progress has been impressive. Although no AI has yet to discover the Nobel winning, the initial form of this idea is already visible in today’s research labs. In 2024, the Chemistry Nobel Prize honored Alpha Fold Creators, which almost about about about how the protein would add to three dimensions. Humans were rewarded behind the project, but it was the AI who made progress. The moment marked a shift. It showed that machines can work more with help. They can answer questions that have stumped scientists for decades.
Other matters are catching up fast. In Carnegie Milan, researchers developed the Coastrist system that applies LLMS to plan and implement complex chemical reactions with robotic lab gear. In Stanford, the AI model is looking for biological samples in RNA data, which has also eliminated highly experienced researchers.
In a Future House like Startup, teams are creating a modular system of AI agents, designed for each specific scientific tasks, such as experimental data, planning studies, or looking for related papers. They are training the AI system to ask their scientific questions, and are experiencing what AI can do without human guidance.
Certainly, these may be merely a baby’s footsteps, but they suggest some big. AI is no longer accelerating the discovery. It is starting to think of itself, and in some very real ways, it may be enough to change the two basically how the discoveries are made and who makes them.
Nevertheless, even despite all this progress, an independent “AI scientist” is out of reach. Researchers at the Alan Institute for AI recently tested dozens of systems and found that most of the small, clearly described can be eliminated, but almost no one can study completely from the beginning to the end. They can predict samples, follow the reaction, and even write reports, but when asked to design and reasoning through the whole discovery, the success rate falls dramatically. In other words, AI can easily copy the consequences of science than it can create a real thing.
This gap can come to something that no machine has yet acquired: human experience. As the Arizona State University has identified by Kambhampati, AI learns only through data. Human scientists live in the world, which they study, curiosity, intuitive, and sometimes even guide through failure.
“I am very supportive of the claims that AI can accelerate science,” says Kambhampati. “But to say that you do not need human scientists and this machine will not only make the sound of Nobel discovering the sound of a hype.”
Yolanda Gul at the University of Southern California (AN Avary The person who views the year) has spent years thinking about what kind of intelligence will be needed for AI because of the methods of scientists. He believes that there is a key lie in the teaching system to consider their own thinking, not just implementing more data. Nevertheless, she admits that the work in this direction has ended in the background as LLM has stolen the spotlight. “There are many interesting results that you can get with productive AI technique,” he told Nature, but there are many other fields to pay attention to it. “
It cannot be denied that AI is improving in scientific research and development. But even if an AI has drawn any inquiry worthy of prize, the Nobel Committee will face a problem in his hands: The rules still say that the prize should go to a person. Perhaps this is the real question now. It’s not whether a machine can win the Nobel Prize, but are we ready to give it a support when it eventually creates a discovery of its prize?
This article first appeared on our sister’s publication, HPCWIRE.
Relevant
