There’s no denying artificial intelligence has come a long way in recent years, and movies like Ex Machina hint at where it could be just a few years down the line. But we’re not there yet.
A new study indicates even the most advanced AI systems on the planet couldn’t actually pass an 8th-grade science test, meaning the good ol’ human brain still has a little mileage left in the tank. The results come from a recent contest held by the Allen Institute for Artificial Intelligence, led by University of Washington computer science professor Oren Etzioni and funded by Microsoft co-founder Paul Allen.
The institute hosted a contest that invited just under 800 teams of AI researchers to build a system designed to pass an 8th-grade-level science test. The best performers successfully answered approximately 60 percent of the questions, which in most school settings is a failing grade. Beyond that, most of the AIs flat-out flunked.
As Wired notes, the five-month contest is a “reality check” for the state of artificial intelligence, and despite all the amazing advances (and they really are amazing!) in recent years, we still have some ground to make up until a computer can really think like a human.
It’s important to note: Yes, we can build computers capable of beating our human champions at games like chess, but building a machine that can actually comprehend something humans do on a regular basis (i.e. read and comprehend, respond on the fly, etc.) is way more complicated.
We’re on the right track, but it’s still a long road. So don’t freak out about Skynet just yet (but let’s not give up total control of stuff like that autonomous warship, you know, just in case).