Like many chief information officers, IBM CIO Matt Lyteson is cautiously and optimistically approaching artificial intelligence. In a recent online interview, he observed that CIOs can't afford to focus on all the new “it” things, but AI is certainly something other than the trend that goes through. “My approach to AI includes traditional AI, generator AI, agent AI, and business automation. All of these tools can be used in concerts to help you translate IBM's business into a digital operating model,” explains Lyteson. “There are some incredibly new AI features, but one hasn't changed. It's best to take a holistic view rather than simply deploying another tool.”
Lyteson points out that generating AI can affect all business areas in particular. “Out of the box, it can often be a dull tool – a sledgehammer for chisel work.” To get full value from generative AI, he believes new recruits need to focus on specific business areas that are ripe for change.
Initial Project
Lyteson's first AI project was to deploy generative AI trained on the company's own data, deploy it across the employee population, and do so in a safe and reliable way. “We already had a lot of AI-powered assistants and automation, and users were familiar with these with specific tasks and features,” he explains. Though relatively simple and not complicated, this new project represented a significant advancement and provided a general enterprise-wide solution with context-specific generation capabilities.
Matt Liteson
This project helped Lyteson move forward based on three basic goals: “First of all, we wanted to give Ibmers the space to play. We experimented with AI with our own business data in a safe, reliable and secure environment,” he says. This approach was designed to minimize risk. It also helped Lyteson equip employees with a specially functional user interface. “This is built on cultural transformation at IBM, directing employees how AI can increase jobs and showing that this happens every day in every corner of the business.”
This approach also established an important reference point for future expansion, Lyteson notes. “Since then, I've been introducing additional assistants and agents. [yet] This remains a single point of input for employees looking to access the ever-growing knowledge and capabilities they deploy in back-end technologies. ”
Effective Planning
Successful implementation of AI requires thoughtful intentionality at first, says Lyteson. “Our plan began by bringing key stakeholders around the table and zeroing in the crystal's clear objective,” he explains. “We didn't allow mission creep, nor did we jump into complex environments without any particular plans.”
According to Lyteson, the project plan began in November 2023 and was completed within 90 days, with the release of the first iteration occurring. “As technology capabilities continue to evolve, they continue to repeat themselves based on feedback and resolution of specific needs, just like insight,” he notes. “For example, routing between different assistants.”
Expectations were met
In many ways, Lyteson said the project is to showcase the incredible possibilities of full-width generation AI for IBM employees, as stated by over 200,000 team members who work all kinds of jobs around the world. “This is an important first step towards making AI an enabler, and we are currently deploying more and more AI-powered agents and business automation across our enterprise.”
“Thanks to our upfront commitment to intentionality, our project met our initial expectations and was able to continue learning and iterate since then,” Liteson says. “The job of a CIO is especially team sports at large companies.”
Lyteson points out that it is also interesting to see how generative AI often spews nondeterministic answers when some business functions absolutely require a deterministic response. “This was an important learning for us,” he says. “As a result, we have taken steps to ensure that the various user bases understand it. [the concern] – Proactively communicate, provide educational sessions, and most importantly embed the appropriate message in the user interface itself. ”
The final thought
According to Lyteson, a single project like this is the CIO's table stakes. “Learning to launch AI projects, experiment quickly and leverage data learners is key to continuing success,” he says. “This was a great example of all these elements getting caught up in a single package.”
