Create An AI Center Of Excellence To Kickstart And Sustain Success

(Published on Forbes.com)

As with all advanced technologies, the testing and adoption of artificial intelligence (AI) has started as a bottom-up (smart people testing new technologies that will help them do their job) and top-down (smart executives recognizing game-changing potential) phenomenon. The same pattern held during the dawn of the internet 20 years ago.

Generating initial momentum with top-down and bottom-up experimentation and strategizing is important, to be sure. However, for AI to deliver actual sustained business value, the enthusiasm of the “top” and the “bottom” needs to be translated into the more challenging part of any organization—the “middle” where any transformation will succeed or fail.

What I have seen work time and time again is a center of excellence, or COE, to coordinate and drive adoption of sustainable AI benefits for the entire organization. The COE should be staffed with a cross-functional team of professionals from multiple levels (including substantial representation from the “middle”), have a clear mandate from senior leadership and be accountable for the important work of exploring, vetting, testing and scaling generative and predictive AI. The key pillars of the COE’s work in my view are:

1. 1. Drive the initial strategic plan: The COE should have adequate outside or internal resources to identify where AI can make the most meaningful impact over the next 3–5 years, the financial modeling of investment and return, the operational requirements of any scaled AI initiative and the KPIs that will indicate success or failure. Strategic planning responsibility should ultimately be ceded back to the business, but lending focus and accountability in the early days will be critical.
      
      
   2. Coordinate the efforts of multiple business units: AI is by its nature a discipline that will touch multiple parts of the organization—sales, service, marketing, operations, IT, legal and so on—and the COE will be a home for the critical tasks of coordination so that any pilots can be fully realized
      
      
   3. Run pilots: A central part of any strategic plan will be piloting the most promising approaches as dictated by the strategic plan. Pilots will need to show the path to scale and as such will need to be well-funded, be cross-departmental and have KPIs that indicate the potential for long-term success.
      
      
   4. Provide a catalyst for developing AI talent: AI is emerging as a “skill” but is only now becoming a “career track”—successful AI professionals will be hard to find on the outside but can successfully be grown if there’s an outlet for their ambition. The COE is the best place to activate the team’s interest and support HR in outlining the long-term requirements for AI talent.
      
      
   5. Assure clear accountability for foundational requirements such as data privacy and security: AI is such a complex topic that ethics, security, privacy, data readiness and many other topics need to be addressed as a part of the initial plan and pilots. In my experience, it’s better to slow down the pilots and deal with these foundational topics versus hurrying to get pilots to market and reversing field later.
      
      
   6. Repatriate the plan back to the business: Just as with digital, what started as experiments are now embedded parts of the business. It may take a few years, but having the COE disband at a date certain will force the cancellation or adoption of the right AI initiatives.

The thoughtful creation of a cross-functional COE with executive sponsorship, resources, cross-functional leadership and an expiration date will help put any organization on a path to realizing the transformational potential of AI.