Siarhei - stock.adobe.com
Artificial intelligence is touted as a set of technologies that will change how every business does business. But, for right now, AI is long on promise and short on proven strategy.
Like digital transformation before it, establishing an enterprise artificial intelligence strategy is so new, there is no playbook to crib from. And even if a playbook did exist, CIOs would still have to contend with the ongoing proliferation of technology, meaning they'd have to customize their investments in AI and plans of action as AI tools evolve.
The good news is that early adopters of AI aren't operating in a vacuum. They can rely on tried-and-true methods for building a new technology competency such as creating a center of excellence. By establishing an artificial intelligence center of excellence (CoE), companies can formalize their vision for and approach to AI.
The case for a CoE
Milind Wagle, global CIO at data center provider Equinix Inc., is just beginning to make a case for an enterprise artificial intelligence strategy. He's tapped members of his enterprise architecture team to keep tabs on the market and to uncover potential use cases. And his team has also taken on a couple of projects to prove out the value of artificial intelligence investments.
One such project is an internal chatbot embedded into the corporate intranet that helps with things like employee training, submitting expense reports and ordering new equipment. "We call it Eva," Wagle said. "And it's a conversational interface that lets the employee community get answers to a whole bunch of questions in a very context-rich manner."
But these efforts are new and, while Wagle is optimistic about where artificial intelligence is headed, he's found that the technologies for use in the enterprise are still immature. Right now, he's busy setting the groundwork.
"As these technologies mature, as we start to get more traction, we are going to put some investment behind this in terms of resourcing and governance and create a team around it," he said. "It's like any other IT capability that you launch: If you're getting traction, then there's a need to nurture, enhance and support it -- and that will require investment and effort."
Artificial intelligence center of excellence
Wagle doesn't call this team an artificial intelligence center of excellence, but he could. Centers of excellence are often established to deal with new technologies, skills or disciplines that don't fit neatly into the enterprise. Their aim is to provide governance and to prioritize efforts. And they can help companies avoid classic pitfalls such as a tendency to focus on technology rather than use case.
"For most enterprises, it's very important that they start from the business perspective," said Bern Elliot, an analyst at Gartner. "The [artificial intelligence] center of excellence is a way to help address the technical needs while also balancing it with the business needs."
Walking the line between shiny, new tech and what the business really needs is especially delicate for artificial intelligence; it's the advances in technology that are propelling the conversation in the first place, encouraging companies to focus on possibility rather than concrete business problems. But, Elliot said, that's a big mistake.
Bern Elliotanalyst, Gartner
"No one does anything with a neural network," he said, referring to the deep learning technique modeled after the hierarchical arrangement of neurons in biological sensory systems. "It's really the applications of AI that make AI useful."
Three parts of an AI strategy
Rather than a single artificial intelligence center of excellence, Elliot teased out three strategic components to consider -- AI business innovation, AI policies and governance, and an AI skills center of excellence. CIOs need to keep tabs on all three, he said.
The business innovation piece is conducted in consultation with the lines of business to determine what applications and uses cases are most important and how they should be prioritized.
The AI policies and governance piece establishes guidelines on what to watch out for and best practices to follow. "You will be using data extensively, in some cases," he said. "In others, you may be exposing the application to customers, and you want to make sure it's done in the right way."
Once a project has been selected, companies will need to get their skills house in order. That's where the skills center of excellence comes into play. The skills CoE can include general skills such as data science and data analytics skills as well as AI-specific skills such as deep neural networks and natural language understanding. "The artificial intelligence skills center of excellence is a way to help marshal and manage your skills," he said.
In many cases, a skills team member won't be reassigned to the artificial intelligence skills center of excellence permanently; instead, team members will participate virtually, tapped to help with a project based on how their particular skills apply.
Finally, the skills CoE doesn't have to be internal employees only. "In some cases, these are difficult skills to develop," Elliot said. Or it may not be in a company's best interest to hire and train employees who leave a couple of years later because their market value is significantly higher.
"This is a perennial problem when a new technology comes along," he said. "It's why companies go to consultants because they get a skills solution provider who comes in and helps complete a project. The company can maintain the relationship, but they don't have to retain the staff."
Virtual CoE is key
Moiz Kohari, senior vice president and chief technology architect at State Street Corp. in Boston, said being fully virtual is key to an artificial intelligence strategy. In fact, he's pushing back against an artificial intelligence center of excellence precisely because it conjures up the idea of a physical group.
As a global company with talent scattered from China to Boston and everywhere in between, Kohari said he finds the physicality of a traditional CoE to be shortsighted. "We find it much more useful to be able to direct the right minds to the right problems," he said.
His model is inspired, in part, by the open source community. Experts from around the world don't have to be working out of the same room on a project like the Linux kernel, a Unix-like operating system. "The way it works is you have maintainers of certain areas of the project and you have contributors to that project," he said. Contributors submit code to a GitHub-like environment where it's reviewed and accepted before it's approved by project maintainers. That kind of workflow exists at State Street.
By giving up the idea of a physical CoE, State Street can think about leadership on a global rather than local scale. "We find that hiring the best minds, wherever they may be, is the most beneficial approach for us," Kohari said.
The company has identified specific technology areas it believes are critical to the financial institution's future, including anomaly detection, natural language processing and predictive analytics. For each technology, they've sought to create a "nucleus" -- an expert or a couple of experts who can build a team and oversee projects that address business challenges in these areas.
These nuclei are "absolute rock stars" in their fields. "I'm not just saying that," Kohari said. "These are guys who aren't only contributing to projects, but started or were founders of some of those projects," he said.
State Street supports their nuclei by embedding management talent onto the team to keep track of a project's progress. "The last thing you want to do is take one of the smartest people who works on anomaly libraries for anomaly detection and burden them with management challenges," he said.