Here are some lessons from former Levi’s and Vodafone chief AI officer
Katia Walsh has been a chief AI officer since 2015 when she joined British telecoms giant Vodafone’s senior leadership team. In 2019 she was appointed CAIO at Levi Strauss and Co, where she remained for four years. Today, she is the chief digital officer and a member of the executive leadership team at Harvard Business School.
Being a chief AI officer involves more than leading business transformation with fast-evolving technology. Due to the nascency of the tech, there is also a need to prove to other executives that it warrants a standalone position. And that’s something Walsh has first-hand experience in.
We asked Walsh what lessons she has gained from her 20 years experience, for other CAIOs or those thinking of taking on those roles, as well as employers considering whether or not to appoint them this year.
#1 ‘Leapfrog into not having done this at all, to going all the way into the latest applications of AI’
Walsh wasn’t hired as a CAIO officer at Vodafone, but a data analytics chief. However, it wasn’t long before she realized that to scale the company’s smart use of data she needed to leverage AI. She therefore advocated for her role to become CAIO – a move she would later replicate with Levi’s when she pushed to include AI in her title.
“As soon as I got to Vodafone, it was really about harnessing data, creating a data repository, and deploying it to business outcomes,” said Walsh. “To get to these outcomes it was machine learning and AI. It was natural and logical for me, but I had to start from scratch.”
At the time, her role was pioneering. During the next four years at Vodafone, she helped build 500-strong data scientist and engineer teams around the world, trained 100 internal associates in machine learning, introduced a governance structure, and received numerous awards, including being named among the top 10 business AI influencers by the U.K. government.
Walsh said there are still plenty of companies currently recruiting for data analytics people and trying to play catch up. But, she sees how automated data analytics is becoming, which makes it one of the applications of AI, rather than a separate discipline.
“It did make sense 10 years ago because companies didn’t understand the value of data and the value of using data to make decisions but they stopped there,” said Walsh. “There are companies that are still trying to adopt this and start there. At this time, we can leapfrog into not having done this at all, to going all the way into the latest applications of AI, which would include analytics.”
#2 ‘If it’s about AI, it cannot be for the sake of AI’
The arrival of generative AI reignited hype around the tech’s entire potential. Figuring out how to incorporate AI into organizations is now a hot topic once again, among companies of all sizes, across industries, and geographies.
Walsh has one solid rule of thumb: “If it’s about AI, it cannot be for the sake of AI. Just like any technology, including emerging technologies, it really has to be about ‘to what end?’”
She says that applying AI in business has to be toward a certain goal and in service of an overall strategy. That means helping AI retain its relevance and gain further differentiation.
“If it’s only ‘oh, I’m here because AI is cool and everyone is talking about it and my competitor is doing that,’ that’s not enough,” said Walsh. “That will fizzle very quickly. Someone will try to hire an expensive talent, but if they don’t know how to deploy that, it will ultimately not succeed.”
Once a company knows that they want a chief AI officer, Walsh suggests that individuals who apply for the job ask themselves why exactly they are interested.
“It cannot be just because it’s cool,” said Walsh. “They need to have a clear vision that is in service of the organization and the organization’s vision, strategy and people, whether they are consumers, businesses, employees, shareholders.”
Walsh is clear that this is a tough job, even in a tech company, because there is still a lot of confusion and a lack of understanding around the technology.
“Most people tend to be intimidated by technology, so you have to really have the knack and talent to humanize it,” said Walsh. “You have to educate a great deal and you have to make sure you’re always connected to some kind of value, whether it’s revenue, cost savings, educational, reach, scale, or helping people get better careers.”
#3 ‘It’s helpful to have the connections within a technical network’
For any job, having a network of mentors is helpful. But when it’s such a new role and technology, it is especially important.
“For this kind of role, because it’s a blend of technical expertise as well as business applications, it’s helpful to have the connections within a technical network,” said Walsh. “It’s also helpful to have connections within a business network. It’s a combination of both.”
She realized that during her time at Vodafone – surrounding herself with people she could talk to, and bounce off, about strategy proved critical while spearheading such a new role.
“This kind of job in a legacy organization can be tough,” said Walsh. “You’re driving the transformation in service of something – strategic goals or customer centricity – but you still are driving a transformation that can be tough. That’s why it’s helpful to have a network of people who are experiencing the same challenges and opportunities who you can confide in.”
#4 ‘You do need to break some longstanding rules’
Toes don’t typically like to be stepped on – especially C-suite toes. Adding another executive into a busy executive team that already has a chief data officer, chief technology or information security officer may create tensions as clarity around roles and remits are determined. CAIOs need to be willing to work well across these departments.
“Maybe you do need a chief data officer, but data is always an enabler,” said Walsh. “I believe that a chief data officer can report directly to the chief AI officer. If they are peers, it’s not going to be clear, because all of what the chief AI officer has to do is predicated on data. Unless there’s a very, very protected and productive relationship, it will be hard to make it work.”
However, working with the chief information security officer is “non-negotiable,” she said. “It’s such an important area. You cannot protect your customers if you’re not protecting the data,” she added. This needs to be a solid partnership to avoid AI and data being used by bad actors, which will come from security policies and practices that have to be in place.
“I believe to drive a transformation, you do need to break some longstanding rules because by nature a transformation is disruption, but when it comes to information security, it’s just not negotiable,” added Walsh.