New algorithm aces university math course questions
Machine learning models have consistently struggled with topics like multivariable calculus and differential equations. These are subjects that many MIT students excel at without any effort. Machine learning models are only able to solve elementary and high school math problems, but they do not always provide the right answers.
A multidisciplinary team of researchers, including Iddo Dori, a lecturer at the MIT Department of Electrical Engineering and Computer Science, has developed a neural-network model that can solve university level math problems within seconds.
The model can also generate new problems and explain solutions in university math courses. The researchers presented these questions generated by machines to university students. They were unable for them to distinguish whether they were created by a machine or a person.