Designing an artificial intelligence center of excellence for disciplined transformation

The heart of establishing a center of excellence is to achieve scale, sustain innovation and expand cultural adoption. The pillars of vision, collaboration and culture will help your organization use data faster to realize value.

Are you optimistic about the transformational business outcomes that artificial intelligence (AI) can affect? Do you envision your organization moving from the basics of assisted intelligence with data-driven decisions into the next generation of augmented intelligence, where your organization designs machines to enhance human intelligence and interactions? It’s all possible. But it’s not possible without people. People drive culture.

Set the vision. Enable cross-functional collaboration. Design for culture.

Vision. Collaboration. Culture. These are the pillars required to set the stage for a world-class AI center of excellence. Whether you’re interested in standardization and integration, leveraging data assets, or measuring business value, a center of excellence (COE) can provide the governance or resource balancing necessary for organizational success.

Vision: A lesson in the centralization of value

To truly grasp the vision behind a center of excellence, let’s go back to its roots. Centers of excellence originated way back in the Qin Dynasty of China in 221 BC. The COE is founded on the concept of centralization of authority that captures the systematic and consistent concentration of that authority. The advantages of such centralization include straightforward responsibilities and duties, clear decision-making, and the belief that centralized power more strongly promotes the interest of the individuals who hold that power. Following upon the heels of these advantages are the disadvantages, which include decisions not being made by those resources with the best knowledge of the problem, delayed execution due to slow information dissemination, and exclusion of certain resources that could have contributed to a better decision.

During the Industrial Revolution, we started seeing new manufacturing concepts emerge—specifically, a shift from the Domestic System (Putting-out System) to the Factory System. The Domestic System shipped materials to rural producers, who often worked out of their homes. This decentralized approach provided products of reasonable but inconsistently quality at a relatively high price. The Factory System centralized the production of goods on a massive scale using machines. This new approach decreased costs and increased worker efficiency. The adoption of the Factory System presented a new question: Where should we produce the goods?

Collaboration: The parallels between AI COE structures and location theory

The degree of collaboration between the AI COE and the rest of the organization is directly affected by the AI COE’s organizational design. Location theory concentrates on predicting the ideal geographic location for economic activity. In short, where should a business be located? Do we focus on variance costs? Is the friction of distance between production and distribution a problem? These concepts seem very embedded in the old world of manufacturing—a world that today’s business executives rarely talk about or reference.

Interestingly enough, we can start to find parallels between designing the structure of pioneering COEs and location theory. Three fascinating location theories help us to understand the organizational-design considerations inherent to setting up an AI COE.

  • Alfred Weber, a German economist, suggested that manufacturing plants should be located where costs are the least (least cost theory). The Weber Theory considered such costs as transportation, labor and agglomeration.
  • Harold Hotelling, and American mathematical statistician and an influential economic theorist, developed the Hotelling Theory. This theory introduced the principle that the location of the industry can’t be understood without reference to other industries of the same kind, and he recommended locating near customers (clustering). The idea of locational interdependence connects a business’s location to its ability to operate and generate a profit.
  • August Lösch, a German economist concentrating on regional and urban economics, studied the economics of location. Lösch’s Theory explained that manufacturing plants should be in locations where they can maximize profit (zone of profitability).

Weber’s Theory, Hotelling’s Theory and Lösch’s Theory all generally agree that business should gravitate to where net profit is the greatest. At the surface, this seems like common sense. When applying this concept to COEs, we’re less concerned with profit and more concerned with continuous value delivery (which may or may not immediately involve profit). We now are left to ponder where in our organization is the maximum value generated?

Culture: finding a home for the AI COE

History isn’t a good predictor of future success. Yet, it doesn’t hurt to understand it and take into consideration the mistakes made in the past so they’re not repeated—in original form—by your organization.

We arrive at the following observations:

  • Centralized techniques effectively distribute communications with minimal opportunity for message variation.
  • Decentralized approaches help to generate worker buy-in due to their customized and distributed nature.
  • Location theory introduces the concept that businesses should be located where the value is maximized.

Extrapolating these concepts, let’s apply them to an AI COE. When we teleport from the COE of late to the present, we discover several useful lessons:

  • Centralized management of a COE is most effective for consistently communicating the organizational vision.
  • A decentralized model for employee engagement helps to co-create ownership and accelerate the COE’s organizational adoption.
  • COEs should have functional reporting centralized (hardline), but it must be operationally (day-to-day or soft line) embedded where maximum value is generated (inside the business).

These concepts offer insightful principles when we design and construct our AI center of excellence for durability, resilience, and accelerated adoption. Specifically, intelligent organizational design is crucial for supporting an evolution in organizational values, beliefs, and behaviors to embrace a data-centric culture that supports an AI-first mindset.

Building an AI-first mindset is all about culture and people and less about technology and data. Armed with an AI-first mindset, surrounded by the right people, and supported with an AI culture, is your AI COE able to measure its value?

Measuring the AI COE’s value

Define success for your AI center of excellence by setting the vision. Identify principles that the AI COE will use to make future decisions. What are the expectations of the AI COE? What’s the AI COE’s essential value proposition? How will value be co-created to deepen shared ownership?

Once the vision is defined, shift your focus to how collaboration will be operationalized. How will the AI COE team collaborate with existing internal organizational constructs (teams, departments, and divisions)? When will the AI champions be identified and engaged? How is the AI community incentivized to participate?

Now begin to design for cultural transformation. How is the “data-as-an-asset” culture tipping point defined? What benchmarks show demonstrated culture shifts in behavior? How are individual performance plans linked to an AI-first mindset? Which communication and media forms will be leveraged for education and awareness?

Planning AI workshops, identifying AI pilots, and creating an AI lecture series can all advance the adoption and make AI real for your organization. However, it can be difficult to know when you’re making progress. The following questions can help you evaluate if you’re making a positive impact on value realization with your AI initiative:

  • Are business leaders able to link AI programs to business results?
  • Do existing business champions identify as part of the AI community?
  • Have AI pilots quantitatively advanced value realization?
  • Can organizational resources articulate how they’re contributing to the knowledge of the AI community?
  • Do resources access the organization and explain how their roles have changed in light of an AI-first mindset?
  • Are randomly polled organizational resources able to clarify the value proposition of the enterprise AI COE?
  • Is the AI COE integrated into existing business processes, or does it sit on the bench as an outsider to organizational value creation?

Your technology partners are going to suggest you start the AI COE journey by taking data-asset inventories, reviewing the technology stack, and exploring sharing technologies across siloed areas. These will all be necessary in due time, but the success of an AI COE begins and ends with people, not data. Set the vision. Enable cross-functional collaboration. Design for culture.

Copyright © 2019 IDG Communications, Inc.

7 secrets of successful remote IT teams