Artificial intelligence designs physics experiments that no scientist has ever done before
Posted: Mon Jan 27, 2025 3:44 am
The use of technology for scientific research dates back about 75 years.
But many scientists believe that the latest techniques in machine learning and artificial intelligence represent a fundamentally new approach to science, drastically different from the way we have understood things before.
If we have so far learned about nature through observation and simulation, technology goes one step further. Observation and simulation help scientists generate hypotheses that they can then test with the help of artificial intelligence to prove or disprove their validity.
The answer to one of Einstein's mysteries
One day, quantum physicist Mario Krenn was sitting in a cafe in Vienna, observing the results coming out of Melvin – his machine learning algorithm.
Melvin's task was to combine and align the building blocks of standard quantum experiments and find solutions to new problems.
This algorithm found many interesting solutions, but one experiment that didn't make sense caught Crane's attention. He immediately thought that his program had a bug because there was no solution.
Rather than ignore this apparent failure, Crane delved into the rcs data uk of the experiment, realizing that he had found the answer to one of Albert Einstein's mysteries called spooky action at a distance.
The solution to one of Einstein's dilemmas was completely accidental. Krenn and his colleagues started this entire research program by trying to figure out how to experimentally create quantum states of photons entangled in special ways.
When two photons interact, they become entangled. Both can only be described mathematically using a single shared quantum state.
If you measure the state of one photon, the measurement immediately corrects the state of the other, even if the two are miles apart. This is where Einstein's observation that photons are "scaredy" comes from.
Generative modeling as a new approach in scientific research
In addition to observation and simulation as a way to new inventions, technology brings us an additional approach – generative modeling, which is completely different from these two approaches.
But many scientists believe that the latest techniques in machine learning and artificial intelligence represent a fundamentally new approach to science, drastically different from the way we have understood things before.
If we have so far learned about nature through observation and simulation, technology goes one step further. Observation and simulation help scientists generate hypotheses that they can then test with the help of artificial intelligence to prove or disprove their validity.
The answer to one of Einstein's mysteries
One day, quantum physicist Mario Krenn was sitting in a cafe in Vienna, observing the results coming out of Melvin – his machine learning algorithm.
Melvin's task was to combine and align the building blocks of standard quantum experiments and find solutions to new problems.
This algorithm found many interesting solutions, but one experiment that didn't make sense caught Crane's attention. He immediately thought that his program had a bug because there was no solution.
Rather than ignore this apparent failure, Crane delved into the rcs data uk of the experiment, realizing that he had found the answer to one of Albert Einstein's mysteries called spooky action at a distance.
The solution to one of Einstein's dilemmas was completely accidental. Krenn and his colleagues started this entire research program by trying to figure out how to experimentally create quantum states of photons entangled in special ways.
When two photons interact, they become entangled. Both can only be described mathematically using a single shared quantum state.
If you measure the state of one photon, the measurement immediately corrects the state of the other, even if the two are miles apart. This is where Einstein's observation that photons are "scaredy" comes from.
Generative modeling as a new approach in scientific research
In addition to observation and simulation as a way to new inventions, technology brings us an additional approach – generative modeling, which is completely different from these two approaches.