BACKGROUND IMAGE: iSTOCK/GETTY IMAGES
Gail Evans, chief digital officer at Mercer, met with us at this year's MIT Sloan CIO Symposium to discuss Mercer's multifaceted strategy to develop a digital business model.
In this video, Evans discusses some of the aims and challenges of Mercer's AI program. The HR giant is applying AI to better understand important key performance indicators, such as customer retention and purchase intent. It's also measuring the output of its digital workers -- aka bots. But getting a return on AI is not an overnight thing. One of Evans' big responsibilities, she said, is convincing business partners to "stay the course."
Editor's note: The following transcript has been edited for clarity.
How is Mercer utilizing AI technologies?
Gail Evans: AI means a few things for Mercer. One is that you have to identify the data that makes sense, because AI without data is a nonstarter. So, what we've done is we've created a data platform and started the journey to data monetization: What is the outcome we are trying to achieve with the data? And, from there, we can build an algorithmic economy.
So, how have we done that? A couple of ways. We started with business intelligence, and we've created algorithms for client retention and product predictions: We know from our client intelligence that have X number of products; what is the likelihood of the next product you would purchase?
In addition, we've looked at unstructured data. We are partnering with a third party to start to collect data from unstructured documents and apply AI to that data.
So, we're looking for real uses cases -- real user journeys -- where we can apply AI to deliver business value.
How are you measuring the business value of AI deployments at Mercer?
Evans: We have a few measures in place. We're looking at client retention for our algorithm. We want to predict the clients that are at risk and the revenue associated with those clients, so we track that.
AI also includes process automation. You can definitely use AI to track the digital workforce associated with applying automation to a task.
All of what we do [with AI] is to try to tie it back to a business metric, because at the end of the day, we want to drive revenue; we want to meet the business objectives. Technology is not a drive-by -- it has to be a part of the mainstream business.
Video from the 2018 MIT Sloan CIO Symposium
Gail Evans, chief digital officer at Mercer, talks about the CIO role in digital transformation.
What are some of the challenges you've faced in your efforts to ensure the business value of AI?
Evans: On the technical side, it's the training -- getting the talent you need for AI and getting [the organization] to be able to think differently. If you want to apply AI, you need a subject-matter expert -- the data doesn't just drop in. So, you need a subject-matter expert, and you have to be able to think differently: How do you apply an algorithm? How do you create an algorithm? What is an algorithm, in the case of some employees?
There's a learning process and passion -- stimulating a passion to want to apply AI to deliver business value. It is passion, new talent and being able to create [a] community of folks [to apply AI] -- one community at a time.
Andrew Ng, who helped developed Google's AI program and now runs an organization of his own, has pointed out that AI breaks in new ways. The workflows are different, and part of the IT leader's job is to communicate this idea to business partners.
Evans: With an algorithm, you've got [to] have a control case, and then you apply the treatment, and it's a fine-tuning that happens over time. And so you may not get the immediate benefit, but you want your business partners to stay for the journey. Sometimes, they decide they're not getting enough value, and they'd like to move on to something different. It takes convincing them, 'No, we need to stay the course, but the benefit will come.'
How important is communication in your job?
Evans: It's so important. As a technologist, I speak a different language. You have to find a way to speak about technology through [the business's] lens. What I've done in the past -- and today -- is learn the business and try to apply their terminology to a very complex set of technologies.