Veteran IT leader and early AI adopter Jim Swanson explains how he’s integrated AI and data capabilities with the business to create exponential change. In Jim Swanson’s personal life, AI is omnipresent.
Source: J&J CIO: Embed Data Science Across the Enterprise
“I’m a big biker and runner, and I really enjoy the way AI continually tells me where I can improve my performance,” he says. “I think we all experience AI in different aspects of our lives, from getting help through chatbots online or asking conversational AI devices in our homes about the weather.” But it’s AI’s role in Swanson’s professional life, as CIO and executive vice president at Johnson & Johnson, that likely garners the most interest.
But it’s AI’s role in Swanson’s professional life, as CIO and executive vice president at Johnson & Johnson, that likely garners the most interest. Swanson has overseen a number of AI triumphs and tests during a career spent pursuing leading-edge work. In this “AI From the Front Lines” interview, Swanson spoke with Beena Ammanath, executive director of the Original Postages/deloitte-analytics/articles/advancing-human-ai-collaboration.html?id=us:2el:3dp:wsjspon:awa:WSJCIO:2021:WSJFY21" target="_blank" rel="noreferrer noopener">Deloitte AI Institute, about the benefits and challenges of AI-enabled transformation, the biggest lessons learned as an early AI adopter, unconventional ways to close the AI data talent gap, and why the convergence of AI capabilities and business outcomes is critical.
Ammanath: What are the biggest AI-enabled business opportunities you’ve seen come to fruition in your career?
Swanson: When I worked in the agriculture industry, our mission was feeding the world; the goal is to use and optimize every bit of variable land. We applied AI to improve outcomes for farmers—determining how much fertilizer to use, how much water to provide, what seeds perform best, and what pieces of land are in better condition.
In health care, there are so many opportunities to apply AI: from accelerating research and innovation to better predicting protein structures for disease therapies to better understanding patient behaviors to keep them healthy.
Yes, the opportunities to apply AI in health care are endless. How is J&J organized in terms of AI and data?
The organizational model should reflect the maturity of the AI capabilities. For organizations just getting started with AI, a centralized model typically builds critical mass. Then, it should be distributed. I like a hybrid model. Being fully decentralized is like herding cats—the technology does not get leveraged effectively. At J&J, IT owns the technology that underpins AI: cloud computing, data repositories, APIs. But we’ve embedded the data science in our functions, where it is closest to the business problems we’re looking to solve. We’ve also created a data science council, with representation from each part of the business, that oversees our portfolio, talent, and technology.
I love the council idea, particularly as it allows your team to embed AI throughout the organization. How do you stay plugged into the rapidly evolving AI ecosystem?
We have what we call a digital radar in IT, with areas of focus such as cybersecurity, internet of things, and data science. We keep up with the leaders in the space and work on proofs of concept, tying everything back to business problems to be solved.
We also do employee hackathons. These allow participants from all parts of the business the opportunity to address real business problems, whether that involves using data to attempt to better diagnose interstitial lung disease, analyzing customer sentiment on social media, or improving demand forecasting for contact lenses. These collaborative experiences enable us to expand our data science community, develop talent, and provide engaging experiences for J&J colleagues. In addition, we’re on a mission to develop so-called “bilingual data scientists” who understand their domain and AI.
Talent is absolutely a critical component of any AI program, yet it can sometimes feel like the necessary data science skills are in short supply. How do you handle that challenge?
Well, you’re never going to have as many data scientists as you want. And it’s not just those specializing in data science—it’s data engineering, cloud engineering, and domain expertise. We built a data science microsite, and we’re talking externally about the type of problems that we’re trying to address. That’s one way to attract talent.
Internally, we conduct a data science showcase each year and offer data science rotations. Most importantly, we listen to our data scientists about what excites them, what they want to learn, and how they want to evolve. Those become our guideposts for growing the data science community.
That kind of creativity and agility is key! Do you look for talent in any unconventional places?
You can get some amazing talent by contributing to open source or offering problems to solve in online communities focused on data science. We welcome individuals with nontraditional backgrounds. People from mathematically driven disciplines—that might mean chemistry or music majors—can be perfect for data science or engineering. We have also—no surprise here—applied AI to the problem, using a skills inference engine to unearth employees with data science aptitudes.
You really need a full spectrum of capabilities to take AI solutions to market. Then there’s what happens after deployment. Our latest “State of AI in the Enterprise” survey points out that awareness of AI’s risks has grown along with its usage. How do you stay on top of those?
I’m part of a cross-industry data consortium, and ethics is one of the issues we study. Having unbiased data to feed into our models is very important. Our leadership team has been talking about data ethics and policy, which is especially important as we look at deep learning and quantum computing.
[wsj-responsive-pullquote html=”I could develop all the models in the world, but if no one is using them, there’s no value. You’ve got to work with the people who want to evolve.”
What other advice would you offer leaders in the AI space?
Think of data as an asset. Very few companies are embracing data as the core of insight and decision-making. That requires spending time to understand where you’re at with data and where you want to go.
Also, never underestimate the need for change management in increasing the use of AI. I could develop all the models in the world, but if no one is using them, there’s no value. You’ve got to work with the people who want to evolve. Let them be amplifiers for what you’re doing and bring others along. If you focus on the detractors, you might get the model right, but you’re never going to move the needle on adoption.
I’ve found that if you provide some AI fluency, there is greater openness to adoption.
I agree. With 135,000 employees at J&J and just a couple thousand data scientists, we can’t do it all. So we’ve introduced a digital learning academy to demystify data science for colleagues across the enterprise. We’re keenly focused on building digital acumen outside the technology organization—how we can take a supply chain engineer, a procurement professional, and a finance person, for instance, and make each of them more digitally proficient.
You speak a lot about the positive impact of the work your organization is doing in the AI space. How can the rest of the C-suite become boosters for AI achievements?
I tell my own organization, “If I am the only one speaking about AI, who cares?” I can influence the 5,000 people in my organization, but not the other 130,000. But our chairman can. Our vice chairman can. Our head of pharma can. And all of our advocates within our data science council and the broader data science community can. That’s why we embed the data science strategy with business strategy. It can’t be two separate conversations; it has to be integrated. We have worked hard over the last couple of years to create that convergence of digital capabilities with business outcomes.
Even amid this growing AI awareness, myths often persist. What is the biggest myth you still hear about AI?
“If data science use expands, my job will be at risk.” And this is simply not the case. AI is an enabler to unlock the thought leadership of every employee from a shop floor line worker to a commercial leader to a junior finance analyst.
In Deloitte’s “2021 Tech Trends” report, we predicted that companies will need to re-engineer their data management value chains in order for their machine learning strategies to succeed. What’s your biggest data challenge?
The challenge in health care is not that we don’t have data; it is making sure we can get approval to access, correlate, and curate that data for insight and decision-making. We are in the middle of a major effort to pull it all together, minimize bias, and apply that data to our most challenging problems. We’re collaborating with hospitals, agencies, insurers, payers, and providers to figure out how to aggregate information so that patients benefit. It is challenging in health care because it is a disaggregated universe. But if we continue to put the patient at the center, it can be done.
Yes, every industry experiences AI challenges but also AI triumphs. What have you learned during your AI journey?
It’s been a journey for sure! When it comes to AI, the model is never right the first time. You’re always iterating, so you need to have a test-and-learn mindset. The question you start with may not be the most important one. As you gain more insight, you can start asking all new questions. I’ve also learned that you need to ask questions people care about. Good modeling doesn’t matter unless it’s solving a business problem.
What’s the one thing you wish you knew before you started your AI journey?
I wish I had known about AI sooner! I started as a scientist and moved to IT because I love the convergence of science and technology. Data science has been around for 50 years; it’s the increase in computing power and data access that has made it more accessible. Getting to this thinking earlier would have been great, but I am pleased with where we are. I am never satisfied because I truly believe this is just the tip of the iceberg. Every year, we learn so much more than the year before.
Editor’s note: This is the fourth article in the series, “AI From the Front Lines,” which goes beyond the hype to reveal the opportunities and challenges enterprises can realize through AI and data analytics, featuring the real-world experiences of enterprise executives in conversation with Beena Ammanath, executive director of Deloitte’s AI Institute.