Five best practice for driving value from artificial intelligence

26 May 2021 Consultancy.uk 4 min. read
Profile

Cognizant has conducted in-depth research on the return on investment of artificial intelligence, interviewing leaders from over 1,200 companies in fifteen countries. Based on the findings, experts from the global technology consulting firm have distilled five best practice for driving value from artificial intelligence investments. 

Pilot and learn

Begin with pilots, but then scale artificial intelligence (AI) applications across the enterprise. Companies starting out should focus on working closely with business teams to identify use cases and demonstrate their value through pilots. It’s important to identify multiple use cases, since some AI initiatives will fail. 

Once pilots succeed, it’s essential to follow through. The real value of artificial intelligence is not in the models themselves, but in a company’s ability to scale them across their organisations. It’s telling that 75% of organisations with high ROI have scaled AI across businesses units. 

The ROI of AI across functions

Hybrid approach

Use a hybrid organisational structure to scale AI initiatives. Beginners often start out with a centralised approach to AI, with a core of data scientists. But these efforts struggle because the teams are often not sponsored by the business lines, which are the ones with many of the ideas. These central service teams are slow and ultimately collapse under their own weight. 

Business people, on the other hand, tend to work in a decentralised way. AI teams need to be close to them, as well as the HR leader, the marketing leader, the supply chain leader, the operations leaders. AI should be seen as a service to them, not something that’s centrally controlled.

For example, Cognizant recently worked with a company that realised its supply chain predictive models didn’t work anymore due to Covid-19. They immediately put data scientists in with the supply chain team and deployed new models in just two weeks. The models went into production quickly because they were tied to a business outcome and the people responsible for those outcomes.

Choosing use cases

Once the organisation grows its AI maturity, it can start establishing standards. How do you know when you’re using responsible AI? How do you eliminate bias? What tool sets are appropriate? How do you integrate third-party data? Which partners do you need? These types of decisions are better served centrally but executed locally as you scale.

Data

Get your data right. Nine out of 10 AI leaders are advanced in data modernisation. That’s why 35% of beginners and 74% of implementers plan to have sophisticated data modernisation systems in place by 2023. Ensuring your data is in good shape isn’t enough; businesses should also bring in a richer set of data, such as psychographic, geospatial and real-time data, which drives higher AI performance.

Industries expecting to accelerate their AI maturity will also surge in data modernization

At the same time, organisations should integrate fast-growing data formats into their artificial intelligence (and machine learning) applications, such as high-dimensional, video, audio and image.

People

Solve the human side of the equation. AI is not just about technology; it’s also about people. Tellingly, artificial intelligence leaders spend 27% of their AI budget on people, almost twice the percentage that AI beginners and implementers spend. It’s critical to hire AI talent that can understand business needs and create solutions, not just build models. 

Eighty-three percent of businesses in the study with high ROI have been successful at developing and acquiring the right people. It’s also important to consider other people issues when adopting AI. Before scaling projects, businesses should put an HR plan in place to address jobs that may be disrupted. 

With AI maturity comes a shift in spending toward people

Culture

Adopt a culture of collaboration and learning. About 85% of businesses that generate large AI returns ensure close collaboration between AI experts and business teams. Also, 83% of high performers are advanced at developing and acquiring AI talent, and nearly nine out of 10 excel at providing non-data scientists with the skills and tools to use AI on their own.

Leaders in artificial intelligence are also more likely to have chief AI and analytics officers in place and multiple executives working together to lead AI initiatives.