Meet Praadhyumn Indaana, 17, whose AI model won him $50,000

Praadhyumn Indaana, a 17-year-old from Montvale, New Jersey, has developed an AI model that could significantly improve safety measures in nuclear reactors. His project, which received the Davidson Fellows Scholarship worth $50,000, utilized a physics-regularized neural network to enhance predictions related to critical heat flux – a crucial factor in preventing overheating within the reactor.

Indaana's model achieved a remarkable reduction in prediction error from around 63% to just 5.5%, surpassing conventional methods. This improvement was made possible by integrating physical theory constraints into the neural network, which allowed it to account for conditions outside its original training data set. The conventional formulas used by engineers in the past had an average error rate of approximately 63%, while Indaana's AI model reduced this significantly to 5.5%.

While this advancement does not indicate that reactors can be operated closer to their heat limit, it could help engineers pinpoint the boundary more accurately and potentially reduce unnecessary conservatism in reactor design and operation. Indaana's work shows how AI can be harnessed not only for data processing but also for integrating scientific principles with machine learning, resulting in more accurate predictions.