HITS

IBM Exec: Quantum Computing to Help AI Grow

NEW YORK – The significant progress that’s been seen in artificial intelligence (AI) over the past few years has been largely enabled by modern advancements in computing — and one innovation that stands to lead computing and AI technology into the future is quantum computing, according to Dario Gil, VP of AI and the IBM Q commercial quantum computing program at IBM.

“It would be hard to overstate the importance that computation has had on the current advances in AI,” he said May 2 during a keynote at the Artificial Intelligence (AI) Conference in New York. He went on to provide a “roadmap” of what he said is “likely to be in front of us” for AI.

“Basically, what we are seeing is [a] two-and-a-half x improvement per year in the performance of the hardware that we can utilize” for deep learning models, he told the conference. Once we move past 2020, “new physics and new devices are going to be required to keep this progress” going, he projected, adding: “Just architectural innovations will not be enough.”

Quantum computing will start to play an increasingly important role in advancing AI, he predicted, noting we’re already seeing an “intersection” between quantum computing and AI.

There is a “special property” in a quantum AI network and “the dimensions inside” of that network “would be exponentially costly to model” with merely a standard computer, he said.

That property is “something called quantum entanglement, and that is really the key resource that is available to us in a quantum computer that does not exist classically,” he pointed out. In a nutshell, using that entanglement can significantly improve classification accuracy, he explained.

IBM has “three working quantum computers available for free on the IBM Cloud,” he said. He showed attendees the significant number of experiments that were conducted using those computers globally March 16-21 and added there have been “over 80,000 users,” while more than 4 million experiments have been run and over 17 papers have been written about those experiments.

In a session at the conference May 1, David Martin, IBM cloud evangelist, explored cognitive functionality in conjunction with edge computing and Internet of Things (IoT) sensors and actuators for eldercare scenarios—specifically the identification of individuals, daily activity monitoring, and aberration detection performed on-premises using HomeAssistant, the Intu open source project and IBM’s Watson cognitive services as part of an experiment he conducted.

The Intu open source project, initiated at IBM, provides a set of service gateways to IBM Watson’s portfolio of AAS cognitive offerings, including conversation and visual recognition.

Martin demonstrated Watson integration with home automation platform HomeAssistant using Intu.

One major challenge with IoT today is “we have somewhere around 10 billion devices and somewhere around 10 million people taking care of those 10 billion devices,” Martin told the conference.

When we get to more than 100 billion devices, as has been projected for 2020, “we’re going to go have to hire another 100 million people to take care of them,” he said, adding: “I don’t know how all of you feel. But I have over 150 electronic devices connected to my LAN at home and it’s gotten so bad that I can’t even remember where they all are.”

Latency, meanwhile, continues to be a challenge as well, he said, noting “latency kills” when it comes to at least certain cognitive applications, such as autonomous vehicles.