Connections

Advancements continue to be made in artificial intelligence (AI) and it’s becoming increasingly prevalent across multiple industries, but misconceptions about the technology remain, according to Beth Smith, GM of IBM Watson Data & AI.

“There’s progress” being seen “across the board and there’s more to come across the board” when it comes to AI applications, she said Aug. 14 in a keynote at the KeyBanc Capital Markets Technology Leadership Forum in Vail, Colo.

That’s why IBM continues to invest heavily in AI overall, she said, adding that, despite the advancements that have been made: “We’re in the early stages as an industry of what AI can do.”
One area that IBM is seeing an especially strong rise in AI usage is chatbots/conversational systems across industries, she pointed out.

But the two lingering misconceptions among people about AI are that it’s a “magic” way to transform their businesses and that it’s too difficult and costly to use, she said.

“It’s not magic,” she stressed, adding: “It’s about data, it’s about training, it’s about expertise, it’s about understanding how it’s being used — the workflow it’s in, the process it’s in.”
There’s also “this huge misconception that it is complex, lengthy and expensive to get to any benefit,” she told attendees. But that’s not the case at all, she said, noting that benefits from AI can now be measured “in days and weeks, not years.”

On a positive note, however, she added: “I think we’re making progress” when it comes to turning around both of those misconceptions.

Going forward, it will be important to not only enable people to “take full advantage” of the knowledge that can be provided by AI, but also use it with emerging technologies including blockchain, she said.

While discussing the state of AI, she noted there are three main forms of AI: narrow AI, which is where we still are for the most part; broad AI, which involves “learning from one task and applying it to a very different task … and we’re seeing some early signs of that”; and general AI in which machines become capable of general intelligent actions and solving multiple types of problems. Of general AI, she predicted: “That’s for the next century. Maybe they’ll get started in 2050 or later. But that’s where the science fiction will really come in.”

Demand for AI has grown in reaction to the “huge explosion of data” we’re seeing across multiple sectors, she said at the start of the presentation. As just a few examples, she pointed out: “There are 2.6 billion emails every second; 80 million MRIs a year; 600 million forms of malware with 400,00 variations daily; 100,000 new cancer articles a year [and] 70,000 new cybersecurity articles a year.”

It’s important to use AI and machine learning to turn all that data “into knowledge,” she said. But only about 20% of all that data is public, with the rest “inside of enterprises,” she told attendees, adding: “That means it’s behind their firewalls, it’s in the middle of their workflows and business processes, it’s in their supply chains, it’s in how they do their business and how they do their work overall.” Therefore, the “real opportunity” out there is “to unlock that decision support capability” that is expected to reach $2 trillion by 2025, she said.

The opportunity for AI is present across many industries today, but is currently being led by the financial services sector, she noted. In that industry, she said, many banks are trying to figure out how to provide customer service in a timely way for customers, as well as improve satisfaction and reduce the need for as many agents as possible at any given moment — and they’re addressing that via methods that include chatbots.

By 2020, chatbots are expected to account for about $8 billion in value for banks globally, she said.

The IBM Watson AI team is focused on whatever enterprises regularly need to deal with, starting with workflows and business processes, she noted. Data is also typically “not labeled for AI and machine learning systems and so companies need techniques that allow [them] to learn more, to understand more, to have higher accuracy, with less training data being necessary,” so IBM Watson is focused on that also, she said. The company also sees the data that its customers own as a valued asset for them and “in no way do we believe that data is a commodity,” so Watson is focused on techniques that enable that, she said.

When discussing AI today, meanwhile, “a topic that comes up a lot is around trust,” she said, calling that something “we’ve taken very seriously.”

The three main points for IBM Watson are that the purpose of AI is to help its customers and “augment human intelligence”; the value of the insights discovered via AI belong to the owners of the data; and “these systems must be transparent and explainable” to customers, she said.