Four CDOs explain how they are putting AI to work for their businesses

Companies must begin learning new technologies, such as artificial intelligence, through rigorous experiments that safely and reliably explore proven use cases.

That was the conclusion of a group of data leaders in a recent keynote address Snowflake Summit 2024 in San Francisco. CDOs came from four major organizations and explained how their businesses are exploring and using artificial intelligence (AI), machine learning (ML) and generative AI.

Here are the takeaways for other digital leaders from the panel that also discussed key characteristics of successful data managers in a previous diginomica article.

Balance your requirements effectively

Anu Jain, head of data and technology at JP Morgan Chase, says artificial intelligence is “nothing new” in the financial services sector. Her firm has been using this technology for years, especially for risk avoidance and cost-cutting strategies. Recently, the firm changed its focus and focused on making a profit:

Does artificial intelligence play an important role in how to make a digital push at the right time? How do we give people a personalized offer? When we think about a sales person, how can we improve their performance by giving them the right advice?”

Jain says AI can improve intelligent routing in customer contact centers, helping to improve employee experience and productivity. JP Morgan Chase is also interested in generative artificial intelligence and points to the development of software engineering, customer service and operations:

We see how generative artificial intelligence at a certain point helps us to rethink our attitude towards our business processes. But we will always implement thoughtfully and securely, because as a regulated industry we have to think about data protection, ethics and bias.

JP Morgan Chase is focused on building cloud foundations that help the firm develop AI-enabled products safely and reliably. Jane recognizes the excitement around generative AI and says that prioritizing use cases is critical. The firm has a “quad model” where product, data, technology and design work hand in hand to deliver high-quality solutions:

A balance between short-term and long-term needs is most important to us, where product and technology work seamlessly together.

Orientation to the client’s requirements

Thomas Davey, Booking.com’s chief data officer, says his organization is an “active user” of AI, machine learning and generative AI. This activity aims to present products and services to customers at the right time:

We have created a platform to lower the barrier to entry in a safe and compliant way. So, we believe that generative AI is just another tool in the toolbox. We have implemented LLM programs and underlying models into our natural language processing service and imaging platform, making them part of our product delivery strategy.

An example of this work, Davey says, is when Booking.com launched its AI and machine learning travel planner for US customers a year ago. The tool helps companies make personalized recommendations to customers using natural language:

We combine MLs and LLMs into a single interaction with clients. Our strategy for ML and AI is that this technology is at the core of what we do. We believe that this technology will facilitate a personalized and connected experience.

More generally, Davey says the rapid pace of change with the advent of technology seems unique. Booking.com has worked with its product teams to ensure that any implementation of generative AI aligns with the broader business strategy:

We have to be conscious of bringing everything together into a single pipeline, so just because something is labeled as generative AI doesn’t mean it jumps to the top of the queue right away. We are focused on implementing projects that meet the needs of our clients.

Follow business goals

Caitlin Halferty, Ericsson’s global chief data officer, says her company has a long history of investing in artificial intelligence, and some of these initiatives are creating significant business value. An important example is the company’s work on digital duplicates of physical sites:

We can improve efficiency and experiment with new features and functions. We reduce workplace risks by sending fewer people to repair with the wrong equipment. We also have an impact in terms of sustainable development.

Halferty says the digital twins show how Ericsson brings together disparate data from drones, sketches and drawings to achieve important business goals in a standardized way. The company also creates fundamental capabilities that the entire business can use:

So advanced data governance, where we can use artificial intelligence to create data access rules, identify data products that will bring the most value to the business, and see where data degradation might occur.

Halferty says Ericsson has no shortage of demand for projects that use modern technology. While experimentation should be encouraged, new initiatives should align with technology capabilities and business goals:

We are moving to a common data lake, a unified data model, and creating common data inputs across the business. We are fortunate to have a mature use case identification program. It is our responsibility to ensure that these use cases align with business priorities.

Build long-term partnerships

Shahran Haider, deputy chief data officer at NYC Health + Hospitals, says his organization is focused on doing more with less and is at the beginning of its AI journey. Last year, the organization formed a corporate advisory board on artificial intelligence:

As you can imagine, when ChatGPT came out, imaginations were fired and every business owner was getting pitches from different vendors saying, “We can solve this problem with AI.” But not every problem is an AI problem.

The advisory board has three main responsibilities: consider how AI can help the organization achieve its strategic goals; ensuring safety, ethics and accountability for any implementation of AI; determination of return on investment from AI projects. Hyder says the organization has identified high-priority use cases and is allowing people to safely experiment with vendors like Snowflake:

We want to make sure we can ask questions, because that’s when you unlock the potential of data in the enterprise.

Hyder says the key to focusing on the right projects is how technology can help support better healthcare. This approach includes building coalitions and partnerships with other health care organizations that create effective solutions:

We plan to have more conversations like this, as they can help us solve the problems of our local communities by bringing in experiences on a national scale.