How smart can computers get? Not quite human level just yet, but tiny steps are being taken in the right direction. For instance, a system that is known as GeoS actually managed to answer 49% of SAT geometry questions correctly, placing it on par with other average human test-takers. This particular milestone came about as the Allen Institute for Artificial Intelligence, which so happens to be a research organization funded by Microsoft co-founder Paul Allen, showed off their GeoS system.
The Allen Institute mentioned, “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.”
Certainly far from a computer that will be able to take over the world, but to be fair to the entire situation, this is definitely a welcome start, and it opens up a whole new door to a future of amazing possibilities. After all, this does not just deal with detecting patterns in the vein of Siri’s voice recognition and Google’s facial recognition for photos, but it actually makes a decent attempt to figure out just what kind of data to make of it which has been input earlier. Basically, we are looking at an AI that analyzes the diagrams and text it sees, before applying whatever algorithms it has about geometry to know the kind of questions asked.