Tech

Why Artificial Intelligence Is Still Ridiculously Limited

By Donna Bellevue , Apr 03, 2017 06:47 AM EDT
WATCH RELATED VIDEO
LONDON, ENGLAND - FEBRUARY 17: BERTI the robot interacts with a Sony AIBO robot dog at The Science Museum's Antenna Gallery on February 17, 2009 in London. BERTI is a life size humanoid robot built to mimic human gestures. Visitors to the Antenna Gallery will witness BERTI�s gestural abilities and get the chance to participate in live research by telling scientists how they rate BERTI�s performance. <br /> (Photo : Peter Macdiarmid/Getty Images)

Without debate, the current capabilities of artificial intelligence are still so ridiculously limited right now as cited by well-known minds of our century. Mayo Clinic's Chief Information Officer Cris Ross and best-selling author and psychology professor at NYU Gary Marcus both agree that the AI still has a long way to go before it can actually make incredible tech achievements. For both of these seasoned experts, the AI current capabilities can be likened to a thoughtless toddler.

A couple of months ago, Mr. Ross gave an eye-opening endorsement of the technology at the annual HIMSS conference in Orlando, Florida. Despite giving the highest commendation of IBM Watson Health’s AI engine, describing its tangible benefits to healthcare, he also cautioned that the technology still needs time to advance. Furthermore, he cited that the technology is currently "still pretty dumb" and has the capabilities of a "two-year-old".

Although dubbed as the best artificial intelligence today, IBM Watson Health’s AI engine is still driven entirely by semantic models that work only by plowing through data, Mr. Ross explained. According to the MedCity News, it is not yet capable of fully making meaning out of all the data that it is given. In this way, it is similar to a toddler who sees fire and proceeds to touch it.

Similarly, Mr. Marcus remains skeptical of the emerging technology. He believes that it learns the same way how children learn, which means that throwing more data at it won’t necessarily lead to progress in areas such as understanding language. However, he also thinks that there is a very effective way to increase the speed this kind of technology is growing.

According to TechCrunch, the only way to get rid of the limits of artificial intelligence is to approach its development the same way CERN was built: through a vast pool of diverse solutions. Current AI is being worked on by individual and small companies that use small strategies for marketing. If the interdisciplinary, multi-country approaches are utilized, like in CERN, real progress could be made to remove the limits of AI.

© 2017 ITECHPOST, All rights reserved. Do not reproduce without permission.
Real Time Analytics