Before Siri and Alexa, there was Watson.
Appearing as a contestant on "Jeopardy!" made IBM's Watson a household name. But since its debut — and win — in 2011, the computer has morphed into something else entirely: An artificial intelligence tool for business.
The company opened up Watson in the cloud wars, making the technology available on competitors' clouds last month.
Behind the Watson branding are career technologists making the tool work for business customers.
CIO Dive spoke with Beth Smith, General Manager of IBM Watson AI, who in her role has to consider AI's use in the enterprise, the need to create transparency, and the role of ethics in the hyped field.
Starting out as a programmer, Smith has spent 31 years at IBM, surrounded by the software technology aspects of Big Blue her entire career.
She has served in her role three years, drawing on experience managing engineering and product teams of "all stages and all sizes," and understanding of the software product business.
This interview has been lightly edited for clarity and brevity.
CIO DIVE: How have customers changed? What are they asking for now that they might not have been asking for five years ago?
BETH SMITH: First of all, I would tell you, if we were really honest with ourselves, they probably weren't asking for much five years ago because [how] it could be applied to businesses wasn't really appreciated — understood and appreciated — at that point.
I would tell you there's two main points that come up with CIOs. One is, putting it to work in what matters in their business.
That can be anything from customer care for credit cards like what we do with Hyundai Card, to mortgage applications like what we do with Royal Bank of Scotland, to insurance applications like Geico, to HR to agriculture to manufacturing to supply chain. But it's about what businesses do. ...
The second category point is around … how can I trust it, how can I understand it, how do I make sure, particularly if I'm in a regulated industry, that I have what's necessary to meet regulations. But even if I'm not, how do I have a good traceability, auditability, that sort of thing.
How is IBM and your business group working to show AI transparency and help regulated customers alleviate concerns about AI and the black box?
SMITH: That's where we're building in a set of trust and transparency capabilities into the platform so that it's a variety of different models and algorithms that customers may be bringing themselves. ...
We recently released a product that we call Watson OpenScale that is a part of monitoring and managing that running AI decision, that engine that's going to be making those decisions. As a part of it, it has capability in it to detect bias at run time, to automatically mitigate that bias without intruding on the customer's model. It also has capability in it to explain what took place in that black box. ...
This piece of explainability and being able to understand what happened has to be such that the process owner, the business owner, the folks that are not likely in the data science department or in the IT department, can understand what it's telling them about what went into the decision.
What were the real factors that caused that decision to come out because I think that's a key point to enable the scaling of this.
How have you as a business leader promote the Watson celebrity but also try to show what Watson can do beyond just playing a game show?
SMITH: "Jeopardy!" was a way for us to help many people inside the field and outside the field understand the advancements that existed in the field in the technology in 2011. ...
The focus that I've brought to Watson and continue to bring to Watson is how to make it such that it's ready for customer business problems or opportunities….
To me going forward Watson is the celebrity, really, when your customers get their business outcome from it. And in many ways now, yes the brand, we communicate about the brand, but many of our customers really like the fact that they can take advantage of Watson and do it within their brand.
For example, Royal Bank of Scotland, they have a number of chatbot systems to do anything from — I think they have nine different ones — anything from contact centers to mortgage applications, to supply chain, to a number of things there. And their contact center one is named Cora, their mortgage one is named Marge.
It's a way for their brand to come forward and those become the celebrities in their world that then their customers begin to personify and their employees do as well.
Where do you see the role of ethics in development in AI?
SMITH: It's a fundamental part. We even have one of our distinguished research staff members that has a full time job — happens to be another woman — has a full time job to be our ethics AI global leader and that's around influencing internal stuff as well as external and community and a number of different things there. We have it in a number of different teams as well.
Frankly, it's also why we published our cognitive principles probably now two years ago I think. We were the first company to actually publish that. I know more and more others have.
But that came back to this point of the ethics and the transparency and being very deliberate with how AI's being used: how we use it, how it's being used, what we do or how we handle data, etc.
I've heard your CEO Ginni Rometty say that Watson is a "she." Is Watson a "he" or a "she" or an "it?"
SMITH: Let me just be clear with you, you cannot have me contradict [Rometty].
Watson was actually named after our founder, Thomas Watson, and his son, Watson Jr., and Thomas Watson Jr. was the one that really made the decision back in the '60s around the shift to computing and it was a foundational element of what happened to the computing era.
So that's where the name came from.
Correction: In a previous version of this article, Royal Bank of Scotland's mortgage chatbot was misidentified. Its name is Marge.