Interest in Artificial Intelligence (AI) is at fever pitch. When surveyed, more than 80% of executives answer that AI would give them a competitive advantage.
But in reality, there’s a big disconnect.
Only 14% of enterprises have actually deployed AI. (And less than 39% of all companies have an AI strategy.)
It's not surprising really.
Implementing AI successfully is a daunting challenge. It requires leaders to orchestrate seismic and complex change, retrain their existing talent and acquire new expertise.
Which means there are plenty of ways to get it wrong.
I spoke with Jenny Watkins, founder of tech strategy firm anewd.ai, to better understand what leaders need to know to get it right.
Renita: So it seems like AI is all the rage in business. As someone who advises executive management on AI strategy, what do you see happening?
Jenny: Well, we can start by learning from the mistakes companies made with the previous hot trend, digital transformation.
The most fundamental mistake was integrating Talent, Technology, and Tasks into the organization’s structure almost as an afterthought. The focus was on technology first, instead of using technology to underpin a transformation rooted in customer and business needs.
Renita: I see, no real roadmap for transformation.
Jenny: Right. Companies just wanted to hire people from tech companies like Google and Facebook. But you can’t hire or buy innovation. Likewise, you can’t hire or buy AI.
Renita: Looking for the shortcut.
Jenny: Exactly. So I’ve seen companies in the AI era hire a team of data scientists, put them in a data lab and expect big wins. Or buy off-the-shelf vendor models.
That simply won’t work if you don’t have the crucial building blocks in place: 1) a strong data strategy, 2) a prioritization of business problems AI can solve, and 3) embedding of data champions within business lines who can understand the business and the data. The frameworks are outlined in our guide Making Artificial Intelligence Actionable: A Compass for Leaders.
Hiring a team of data scientists is reminiscent of 2015 when many enterprises created a Silicon Valley Lab. Lots of these labs eventually closed because they weren’t providing much value - they weren’t close enough to the business challenges that innovation could solve.
Renita: So where should leaders start?
Jenny: With a hierarchy of their most important business problems. Then they can actually see if AI is the right approach, if their problems are good candidates for AI application: data-rich, error-prone and cost-intensive tasks.
Renita: Got it, not doing AI for AI’s sake, trying to squeeze a square peg into a round hole.
Jenny: Right. So the best first hire is a Data Strategist who understands both the business problems and the data potential. S/he can help design the appropriate teams — which could then include data scientists.
Also, leaders need to up-skill their existing workforce! It's expensive to hire outside talent when you have employees who already know your business and could just be trained on new tools.
Renita: Absolutely. So, where would you put this function?
Jenny: That’s another mistake I see. Often, new initiatives like AI are run on a separate track, not embedded within existing businesses. This is one difference from the Innovation era, where the common theme amongst failed initiatives was lack of executive sponsorship.
In the AI era, the challenge is different - it’s having a 360-degree view: on data, on the customer and on the business. So I recommend putting what I call “AI Swat Teams” in centralized places where they can have a single view on the organization — and where other non-AI team members can learn from them too.
Renita: I see, not taking an ad hoc approach, making sure the execution is intrinsic to your organization.
Jenny: Yes, that’s the greatest risk. At the end of the day, infusing AI into an enterprise is a change management challenge. When you don’t start from within, you risk the initiative running independently on a parallel track and not actually intersecting with your business problems.
Again, you can't just acquire talent and expertise. If you truly want to harness the power of AI, you have to make fundamental changes in your culture, your organizational structure, your learning models — much more than leaders realize. There needs to be a deliberate, thoughtful approach: “What does our business need? What are the problems we're trying to solve?”
Renita: I know data is a big ingredient in this. What mistakes are you seeing there?
Jenny: It’s surprising how many companies don’t have their own data collection strategy. I recommend the COMA approach: thinking about how you are going to Compile, Organize, Mobilize and Analyze data. Of course, there’s a ton of plumbing needed to make that happen - e.g. how does the data feed the rest of the enterprise, create a learning and action loop, etc. And don’t forget governance - especially with privacy and regulation laws like GDPR. This infrastructure work is, in fact, the pre-requisite step to any AI initiative.
Renita: How do the “best in class” view their data strategy?
Jenny: They understand a key point: that data has an expiration date. So they are constantly looking for new sources of data that they can accumulate because they inhabit a particular vertical, health or finance, for example. This “vertical vortex” gives them a moat and, ultimately, a competitive advantage.
A Fortune 50 bank I worked at always embedded data in the product teams. Whenever we were rolling out a new initiative, our core team met frequently and 75% of the people on that core team were data people.
Another best practice is building the tech stack at the beginning. The boldest companies build to scale, so that they don’t get into technical debt later on.
Renita: As always, build with the end in mind.
Jenny: Yes, indeed. Build with the end in mind, integrate Talent, Technology and Tasks early in the process, and start within if you want your AI transformation to succeed.