Title: AI designs physics experiments, and it’s already working.

**Article:** Researchers have taken a major step toward automating scientific discovery. A new system uses artificial intelligence to design physics experiments—and it has already produced results that work in the real world. The project, led by a team of physicists and computer scientists, aims to replace the slow, manual process of trial-and-error in the lab. Instead, the AI analyzes existing data, identifies gaps in knowledge, and proposes specific experiments to fill those gaps. In a recent test, the system designed a set of optical experiments. When scientists built and ran those experiments, the outcomes matched the AI’s predictions. That means the machine did not just guess—it understood the underlying physics well enough to make useful choices. The key innovation is how the AI is trained. It does not simply memorize past results. Instead, it learns the rules of the system, then explores possible setups to find ones that would produce new, informative data. This approach cuts down the number of blind tests researchers must run. The team says the tool is not meant to replace human scientists. Rather, it acts like a smart assistant that suggests where to look next. For fields like quantum optics or material science, where the number of possible experiments is enormous, this could save months or even years of work. The next step is to expand the system to more complex fields, such as chemistry or biology. If it works there, the way research is done could change fundamentally. For now, the message is clear: artificial intelligence is not just for analyzing data anymore. It is starting to decide what data to collect in the first place.