New AI System Can Solve SAT Maths Questions Like Average Human High School Student

Researchers at the Allen Institute for Artificial Intelligence (AI2) have come up with a unique computer system that can read and assess geometry problems just like an average high school students taking the SAT.

The new system, called GeoS, is certainly a groundbreaking achievement in the field of artificial intelligence. It could correctly answer 49% of the all questions it was fed, according to a research paper published on Monday.

“If these results were extrapolated to the entire Math SAT test, the computer roughly achieved an SAT score of 500 (out of 800), the average test score for 2015,” the Allen Institute commented in a statement.

Even though this artificial intelligence is far from taking over the world, but it does have the capability to analyze the information it was given instead of having advances techniques for detecting patterns, like Siri’s voice recognition, Google’s facial recognition for photos etc. In other words a typical advanced artificial computer system could tell whether an image was part of a geometric problem or, say, a bag of cement; but GeoS is actually trying to understand the problem based on the information it has and then uses it, to answer the questions accordingly.

AI2 worked with the University of Washington’s Computer Science Department for the creation of GeoS.

“Unlike the Turing Test, standardized tests such as the SAT provide us today with a way to measure machines ability to reason and to compare its abilities with that of a human. Much of what we understand from text and graphics is not explicitly stated, and requires far more knowledge than we appreciate. Creating a system to successfully take these tests is challenging, and we are proud to achieve these unprecedented results,” commented an ecstatic Oren Etzioni, CEO of AI2.

At present, GeoS can answer plane geometry questions, but researchers hope that it will be solve the entire math section of SAT within the next couple of years.

“We’re excited about GeoS performance on real-world tasks. Our biggest challenge was converting the question to a computer-understandable language. One needs to go beyond standard pattern matching approaches for problems like solving geometry questions that require in-depth understating of text, diagram and reasoning,” commented Ali Farhadi, a senior research manager at AI2.