Driving smarter customer experiences with AI and machine learning

Artificial intelligence (AI) is demonstrating its ability to stimulate growth in both digital and nondigital native businesses. According to Deloitte, businesses across sectors are using AI to create business value. From streamlining data analytics to improving customer experiences, AI offers several benefits for businesses.

When AI is integrated into an organization’s core product or service and business processes, it’s at its most beneficial. Despite AI’s increasing popularity, many businesses still find it difficult to use AI and ML on a larger scale. In a panel discussion during VentureBeat’s Transform 2022 virtual conference, Chris D’Agostino, the global field CTO for Databricks, Patrick Baginski, senior director of Data Science and Data Analytics at McDonald’s, and Errol Koolmeister, AI and data advisor at The AI Framework, discussed how their companies use AI and ML to create smarter customer experiences.

Implementing AI and ML on a larger scale

There is a growing interest in AI, its subfields, and allied disciplines like machine learning (ML) and data science as a result of how AI is transforming every industry and business function. According to a recent McKinsey survey, 56% of organizations are using AI in at least one business function.

Whether as a digital or nondigital native business, Baginski said it is important to always think top line first about the value that can be delivered from AI and ML projects. According to Koolmeister, a 2019 MIT Sloan assessment showed how businesses were having difficulty as they persisted in trying to get their businesses off the ground, noting that  the return on investment  from AI was meager. Koolmeister also cited a recent study conducted by Thomas Davenport and NewVantage Partners that shows that the market has changed — 26% of the world’s largest companies had AI in wide-scale production while 92% of them are currently investing in the technology.

“I think most companies are making some sort of effort in actually implementing AI into their organizations,” said Koolmeister. “There are a few clear distinct things, one of them being building up the internal capabilities in order to be able to deliver on AI large enterprises. You can’t start with a decentralized organization, you need to build central momentum. First, you need to build your first use cases and then grow the maturity while you’re rolling things out into the organization. So it needs to be learned by value and there needs to be clear proof points early on to actually adjust or motivate the investment levels that are required to transform large legacy enterprises.”

Pitfalls of trying to create a robust AI/ML environment

Today’s AI is largely centralized and can be owned by only one entity. This is a significant hurdle for AI, according to Baginski, who noted  that corporations build best practices, standard operating procedures and common platforms for the 80% of work performed by analysts, data scientists and data engineers. However, he asserted that such activities should be viewed as a collective endeavor that fosters remarkable development. 

“I think one of the big challenges is forcing centralization,” Baginski said. “I think there is a reason to say that you’re establishing best practices and common platforms and common processes for the 80% of the work that an analyst or a data scientist or a data engineer does, but you really need to see this as more as a community effort and your success in building out these guidelines is based on the company and the business units adopting it. So forcing centralization is typically very detrimental to that effort.”

Baginski also highlighted another challenge: transitioning from the generalist data science team that handles all the machine learning, data science, measurement, analytics, pipeline creation and so on, towards having several different roles that are more specialized, that each plays a part in the bigger picture of developing a good solution.

“The other challenge is, the devil often lies in the details right? So I think we’ve moved away a little bit from the generalist data science team that is just going to handle all the machine learning all the data science, all the measurement, all the analytics, all the pipeline creation and everything, towards having several different roles that are more specialized, that each plays a part in the bigger picture of developing a good solution,” said Baginski.

Baginski also noted a typical challenge that he has seen is that a company needs to be very clear early on, on a couple of top priorities of projects or use cases that make sense for a team to start off with and that can be used to then essentially derive the adoption of those guidelines into the business units. He added that those use-cases have to be properly vetted by experts for how applicable they are  to the ML and AI idea, how well they serve that and how much value they will drive.

However, D’Agostino honed in on the importance of establishing a team to solve the problems aforementioned.

“You’re not gonna find a unicorn that’s going to solve all these problems magically. There really is a collaborative effort. The business stakeholders are key enablers of getting things done. They understand what use cases need to be driven inside the business,” D’Agostino said.

Baginski said, “in a lot of companies, if you’re serious, and if you’re in a management C-suite, you need to provide training or support without scaling for them to actually be able to drive. So there’s an educational aspect to actually being successful with these things.”

Koolmeister added that consistently educating the workers is absolutely critical, especially if it’s in a large enterprise that is very distributed in many different countries.
Don’t miss the full discussion of what lessons McDonald’s, Databricks and The AI Framework have learned from implementing and scaling large AI initiatives to drive business value and smarter customer experiences.