<img height="1" width="1" style="display:none;" alt="" src="https://dc.ads.linkedin.com/collect/?pid=306561&amp;fmt=gif">

How to Unlock the Full Potential of Enterprise AI

Written by Solomon Radley

How to Unlock the Full Potential of Enterprise AI

Written by Solomon Radley on Mar 23, 2020 1:14:52 PM

Artificial Intelligence & Machine Learning Data & Analytics LIVE 2020

Scaling AI capabilities across an entire organization is the best way to unleash the technology’s true potential to drive ROI

AI will add more than $13 trillion to the global economy over the next decade according to Harvard Business Review. Yet, the majority of organizations struggle to get the most out of their AI initiatives.

The title reports that just 8% of organizations have the right practices in place to support widespread AI adoption. As a result, many AI projects stall before they ever drive significant ROI. Data leaders often struggle to identify the best use cases to trial and scale or are unable to apply proven AI tools to transform their businesses.

“One of the harder pieces that we've solved for is the productionalization and the operationalization of machine learning models. But of course, that’s where the value lies” – Guy Taylor, Head of Data and Data-Driven Intelligence, Nedbank

“We have to change the way the business runs,” argues Dr Alexander Borek, Head of Data, Analytics and AI at Volkswagen. “If it’s not changing the way our business runs, either by changing processes or changing how people make decisions, it’s not worth much.”

Given that the technologies which enable AI in business contexts are advancing rapidly and have never been more affordable, the time seems right for enterprises to capitalize on the technology. But to do so effectively, they must first ensure they have the right data, technology platforms, business processes, organizational structures and change management programs they need to succeed.

“It has to be a team sport,” concludes Matthew Fryer, VP and Chief Data Science Officer at travel company Expedia Group. “It’s data, platforms, use cases and people, and also your best practices, all coming together.”

Laying the Foundations for AI-at-Scale

Data teams need a great deal of technical infrastructure and expertise to successfully deliver AI projects at scale. As such, these initiatives are best suited to organizations that are advanced in their data journeys.

However, many data leaders report that organization-wide demand for AI capabilities is just as important as their team’s ability to deliver them.

“There are two things that we think about when scaling our projects and our products,” says Elizabeth Hollinger, Head of Analytics and BI at generator supplier Aggreko. “The first and most important thing for us is engagement across the business.”

“If it’s not changing the way our business runs, either by changing processes or changing how people make decisions, it’s not worth much” – Dr Alexander Borek, Head of Data, Analytics and AI, Volkswagen

Once Hollinger's team has identified the capabilities the wider business wants, it can assess what needs to be done to put the right processes and technical infrastructure in place to deliver these projects.

“The second thing is that we’ve got the right technology platforms in place to be able to productionize and scale our solution,” adds Hollinger. “We’re undergoing a big program of modernization in our data tech stack at the moment. That will be in place by the end of this year, and that just gives us the ability to scale things up quickly and have them created in a productionized environment that’s easy for us to manage.”

A Scalable Approach to Problem Solving

For an AI tool to deliver maximum ROI, it must also solve business problems in a generalizable way. Due to the amount of resources it takes to develop an AI product, any given tool’s ability to deliver returns will be severely restricted if it can only be applied in a single context.

“For me, scaling is applying a solution to as many value creation points as possible,” says Dr Ahmed Khamassi, VP Data Science at energy company Equinor. “So, we need to approach a problem in a way where we say, ‘Can we build a solution so that the next application costs us next to nothing?’”

Of course, ensuring that AI products solve a business problem end-to-end is a big part of this. If business units require the data team’s assistance in order to make AI tools ‘go’, that will create a long-term drain on resources that inhibits scaling.

“For me, scaling is applying a solution to as many value creation points as possible” – Dr Ahmed Khamassi, VP Data Science, Equinor

“This is a massive paradigm shift, I would say, moving from ‘we just do some modelling’ to ‘we have to solve the business problem end-to-end’,” Dr Borek argues. “We must often build a user interface or integrate into a user interface or make changes to a user interface.”

The Gradual Rise of Enterprise AI

Given the sheer number of moving parts at play, it’s easy to see why so few organizations are successfully harnessing the power of AI-at-scale. Even with all the right foundations in place, many AI pilots and experiments still result in failure.

“I don't think we as enterprises have caught up with this yet,” says Guy Taylor, Head of Data and Data-Driven Intelligence at Nedbank. “There is a massive gap currently, and that gap's been closing a little bit, around the amount of work that's being done to get value out of this stuff and actually getting value out of the stuff.”

That said, digital native start-ups and trailblazing enterprises are already rolling AI products out organization-wide. From Facebook to Amazon, Netflix and beyond, examples of the transformative impact AI can have when applied correctly are easy to find.

Companies that want to share in the $13 trillion Harvard Business Review predicts these capabilities will generate by 2030 would do well to follow the examples of these AI trailblazers.

Related posts