« Why The Time Is Now for High-Performance Cryptography | Main | Rise Of The Digital Insurer »

May 22, 2015

When Will Computers Become Truly Smart?

Posted by Dr. Srinivas Padmanabhuni (View Profile | View All Posts) at 10:51 AM

How smart is today's artificial intelligence? [Source: https://www.youtube.com/watch?v=poLZqn2_dv4]

There's something quite fascinating to the phenomenon of rare, scientific programs reaching the daily parlance of the general population. A current example is artificial intelligence (A.I.). The average person on the street knows what A.I. is and has probably mentioned it in a non-scientific discussion.

I was reminded of this phenomenon when I saw one of the season's first blockbuster movies. It was pure science fiction but the protagonists wrestled with A.I. and indeed battled a monster that learned from humans and independently improved itself. That's always been a concern about A.I. - that the science behind it is somehow sinister. At its worse, A.I. creates a kind of Frankenstein's monster that is out to destroy human life. But perhaps, that's a far-fetched notion.

Robotics and A.I. are technologies that have been with us for a long time. Remember HAL from 2001: A Space Odyssey? That was more than 45 years ago. But the reason the subjects are in the news again is because of just how advanced they are becoming. Our own CEO, Dr. Sikka, spoke about the potential of machine learning and A.I. at a recent conference. There are endless possibilities to this science, but we humans always rest easier when we know that there's a 'kill switch' somewhere nearby.

Up until now, for instance, an unmanned drone aircraft was piloted by a highly trained human pilot on the ground. There are now drones in development by the aerospace industry that fly themselves based on computer programs that continually learn from their surroundings and automatically become updated. That's why we need to start paying attention to A.I. in a big way. Do you want a drone flying overhead that is armed with attack missiles...yet has no human controlling it? Most certainly not.

ImageNet, a database of just over one million pictures, is making a concerted effort to build deep learning (a branch of machine learning) benchmarks and expand the horizons of A.I. Companies such as Google, etc. are competing to build A.I. programs to classify the images on ImageNet, with the lowest error percentage. Microsoft's software has a 4.94% error rate and Google has achieved 4.8%.

Ray Kurzweil, the technologist who ran an A.I. think-tank for 15 years, calls A.I. 'accelerating intelligence,' and has recently predicted that computers will be as independently intelligent as humans or more, by the year 2030. To scientists like Kurzweil, who in 2012 was appointed a director of engineering at Google and heads a team developing machine intelligence and natural language understanding, A.I. is going to radically change the world. It's interesting how he looks at A.I. − not so much as a technology but a concept.

Kurzweil says it's a concept that underlies just about everything that touches us in our day-to-day lives: biotechnology, nanotechnology and materials science, molecular electronics, computation, neuroscience, physics, the Internet, energy, electronics, pattern recognition, virtual reality, human brain reverse engineering, brain and body augmentation, and, yes, artificial intelligence and robotics. These fields have one thing in common: They are extremely dependent on pervasive information-based technologies in order to advance themselves to the next level. So now that we're fully part of the era of Big Data, everyone from a scientist to a corporate director is trying to figure out how to reap benefits from A.I.

An interesting take on the current state of things is from Fei-Fei Li, the director of Stanford University's A.I Lab. She recently said in an interview that we're not as close to achieving A.I. that is on par with human intelligence as some might think. The reason, she said, is because writing such algorithms and designing robotics to house self-improving computer brains are extremely challenging. So a note to all those computer programmers and software developers who fear that their jobs might give way to an A.I.-enabled device sometime soon: take heart - we're not quite there yet. But society in general and enterprises specifically should be preparing for such a day sooner than we expect.

Post a comment

(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)

Please key in the two words you see in the box to validate your identity as an authentic user and reduce spam.

Search InfyTalk

+1 and Like InfyTalk

Subscribe to InfyTalk feed

InfyTalk VBlogs: Watch Now

Infosys on Twitter