Implementing an AI strategy and framework

Publish date:

The journey to an AI-driven future operating model requires continued executive support, a defined business strategy, and solid data governance. Prepare to experiment, and learn from your experiences.

In the first two articles in this short series, I discussed how to build a strategy for an AI-driven business, and how to build a framework for its implementation. In this, the final article in the sequence, we’re considering how to take the strategy, together with the framework, and move towards implementation of an AI-driven business model.

The first point to make about implementation is how important it is to keep everyone pulling in the same direction. Senior support and sponsorship are required, especially as artificial intelligence (AI) can have quite a disruptive effect on the way a business operates. The leadership needs to steer the ship and maintain the course, when there are likely to be crosswinds and counter currents.

It really helps if the organization can develop a community of advocates and champions. Artificial intelligence can fundamentally change working life, by taking away certain typical, human-centric tasks and augmenting the activities that remain. Advocates and champions will be essential in helping to manage the disruption, and in providing any necessary reassurance.

It also helps if the business approaches the transition pragmatically. It’s important to start small, solidify, reflect, scale, and grow.

Pitfalls to avoid

Adopting AI isn’t easy. It demands a different mindset and sometimes is hard to know when you’ve got it right. It’s an innovation that shouldn’t be adopted for its own sake. Without focus, it’s easy to spend every day experimenting, but ultimately not taking the organization forwards.

Another pitfall is the availability of skills. There is a void between the data scientist and those who are less well versed in math. If that void cannot be bridged, there’s a possibility that a great idea will not make it past the experimentation stage, as it lacks backing from within the business. Worse still, the depth of need may not be not fully understood, and the solution ends up being an oversimplification.

A very important, and growing agenda item is to be mindful of ethical considerations. These are essential, especially where AI is being used to make potentially life-changing decisions. For instance, dealing with algorithmic bias is problematic where such biases can be highly consequential. How can cognitive systems be designed so as to avoid, for example, an inherently narrow set of cultural assumptions?

Finally, for now, the rationale of the AI application should be fully considered. It wouldn’t be right to think of AI merely as a form of automation to drive cost savings. An example might be the use of AI “agents” in a call center merely to remove costs. If, instead, the starting point is to use AI to optimize and humanize customer interaction, it’s likely to foster a more positive design thinking approach that will get the best out of the solution. This, in turn, will encourage customers to perform more self-service, ultimately leading to an overall reduction in cost to serve for the call center as well as improving customer satisfaction. In short, both objectives are met, but are derived from a more AI-positive stance.

Watchwords for success

Let’s close this short series of articles with a few key – and business-oriented – points to bear in mind.

First, and as noted at the outset of this article, support from the top of the organization is critical to ensure ongoing focus, especially when the journey to an AI-driven future operating model is likely to have highs and lows.

Next, AI is not a bolt-on. Organizations should aim rather to have a business strategy that recognizes the transformative nature of AI, rather than as an additive to what’s being done today.

Further, organizations should expect to experiment and to experience some early failures. Few companies get AI right first time.

Finally, businesses should develop a manifesto for acceptable and ethical usage of data. They should ensure governance models are in place to control and protect the quality of data. They should also make sure that control mechanisms are in place to test for, identify, and eliminate bias from algorithms.

Artificial intelligence is a developing area. It’s fraught with complexity, and is a topic of serious research, and so the controls should be revised frequently.

Is it a challenge? Indeed it is. But do the potential rewards make it worthwhile? They absolutely do.

Download Capgemini’s TechnoVision 2020: Future Thinking Simplified, a report that helps business executives anticipate and assess emerging technologies as part of their strategy creation.

Read other blogs in this series :

Lee Beardmore  has spent over two decades advising clients on the best strategies for technology adoption. More recently, he has been leading the push in AI and intelligent automation for Capgemini’s Business Services. Lee is a computer scientist by education, a technologist at heart, and has a wealth of cross-industry experience

Related Posts

Business Services (global)

The 80/20 rule – how you can gain more by doing less

Priya Ganesh
Date icon October 8, 2020

To get better results, you often need to do less but focus more. The 80/20 rule can help you...

#gradathon

Customer service – a path to a fulfilling and successful career

Ramesh Balasubramanian
Date icon September 29, 2020

Putting the customer front and center of our business can give you the confidence to provide...

Technovision

Building a framework for AI implementation

Lee Beardmore
Date icon September 11, 2020

Addressing the breadth of AI applicability, the way data is governed, and the outcomes you...