4 goals to target when building AI skills

Having discovered the frightening costs of hiring AI experts from outside the organization, some CIOs have developed ways of building skills in house — not just within the IT department, but across the enterprise. Many of the early movers targeted four distinct methods that should now be considered for AI training programs in any organization.

Increase office productivity

Construction services company Arco set up its first AI training to help people use Microsoft Copilot to transcribe meetings, create action items, and insert them into Microsoft Planner. They started with a limited group of executives and executive assistants, and brought in experts from Microsoft to conduct the training. “Our intention was to make sure people understand how to use the tools and know how to set the privacy so none of the meeting transcripts go outside our company,” says Robin Patra, Arco’s director of data and AI.

The success of the initial training was measured in three ways. The first was to take the ratio of how many times attendees activated Copilot versus how many meetings they attended in the weeks following the training. The second was to count the times Copilot was integrated in a workflow following a meeting. And the third way of evaluating the program was to survey attendees to find out if they were satisfied.

The initial session was followed by a successful pilot in October 2024. Arco then rolled out the tool to the rest of the company in November and training is now obligatory for all 4,000 employees. Patra refers to this course as 101, and says it’s now offered as an online class that takes around five hours and covers the fundamentals of AI.

Improve core functions

After the success of their office productivity course, Arco added a second level of training to help people match AI to business challenges. This course, which Patra refers to as 102, is a five-day online training program that goes through the entire construction process from job estimation and design, to project management and execution. The goal is to get people to think about how AI can improve core processes for construction services. “This second course is voluntary, but we encourage people to take it,” says Patra. “So far, around two-thirds of our company have done so.”

Following the training, attendees of the second course are required to put at least one new idea into the company’s innovation portal. Then the innovation and engineering team debate on those ideas, and in some cases reach out to the people who contributed them to find out more. “For example, one of our legal team members who learned about LLMs submitted an idea,” says Patra. “Whenever legal gets a new case, they have to go through 500 to 600 pages before they can respond. That person pointed out that the new legal case may be similar to something they’ve done before, and they could leverage what they had already done. So we implemented a legal AI that helped find similar cases and provide an initial response.”

Arco went on to implement a third training program for enthusiasts who want to work on AI using low-code and no-code tools to create prompts specific to the construction services business, for example. “The third training is for the people who’d like to get their hands dirty and build something,” adds Patra. “It’s very specific to the people interested in doing low-code and no-code in the AI tools.”

Around 80 people have taken the program for enthusiasts, which is offered every quarter. “We hire outside instructors to teach people new skills,” says Patra. “It’s a hands-on program that can’t be done online, so we book training and bring everybody into our head office in St. Louis.”

Develop AI skills across the organization

Engineering companies are often the quickest to adopt new technologies. Take for example Lexmark, which four years ago partnered with the North Carolina State University, becoming a founding member of what the school calls the AI Academy. And as part of that undertaking, Lexmark can enroll people in degree programs without having to pay tuition, says Vishal Gupta, Lexmark’s CITO.

Four years ago, the company had just five data scientists dedicated to AI. Now it has 100 who’ve completed four key tracks. But the training isn’t limited to any specific group — it was extended to people across the company who don’t necessarily want to become hardcore specialists. So people from HR, finance, manufacturing, and other lines of business who have a desire to learn and invest in themselves enroll, and even if they have no programming background, they’re taught Python to get a feel for how to create an AI application.

Volunteers from within the company attend three-hour classes four nights a week for a year, and are assigned mentors within Lexmark and given projects, which not only complement the programs, but target business objectives within the company. According to Gupta, nobody has dropped out of the course and very few participants have left the company. “People are happy to have the opportunity to invest in themselves and apply what they learn,” he says.

So far, six cohorts of people have gone through Lexmark’s training program, helping the company not only develop a talent base, but find the use cases for AI. “After somebody from manufacturing, customer service, sales, or any other business area completes the course, they know exactly which problems they can bring to us to solve with AI,” says Gupta.

Build an AI culture

According to Marc Booker, vice provost of strategy at the University of Phoenix, sometimes the best way to learn technology is through practical hands-on experience. To this end, the university established communities of practice, or mentorships, where technologists who already know how to use a model help solve problems in the organization with colleagues who are less practiced but eager to learn.

According to Booker, a community of practice is an open forum where people with similar interests can share ideas. Groups typically have goals, such as developing skills in machine learning or LLMs, and as much as possible, Booker encourages leaders within the university and partner organizations to build teams with a mix of experience. “Take for example, a case where you want to implement an LLM bot to improve customer services,” he says. “Put an open call out to the community of practice to find out who’s interested in the problem you’re solving and who may have done something similar before.”

According to Booker, creating communities gets you in on the ground floor for change management. “You build their skills and you minimize their fears of change,” he says. “Once you build these communities, you’ll be surprised by which people shine and move up within the community.” But the user groups do need organization. Somebody should be designated to play the role of facilitator at a minimum — and some community members will naturally develop into mentors. Booker says people from the business side sometimes morph into technologists.

Lexmark also wanted to go beyond just training people on building AI applications to help users lose their apprehensions about AI and become early adopters. To help build a stronger general culture within the company, last year Lexmark launched a course called AI Foundations. “We thought maybe a thousand out of our 7,000 people would sign up,” says Gupta. “It turned out that 5,000 people took it within two months.”

Gupta isn’t the only one who’s been impressed. Forward thinking CIOs and IT leaders see enthusiasm for AI as an opportunity to go beyond just implementing new tools. They design training programs that foster a culture of innovation and empower employees to solve problems in new ways. By focusing on the right set of outcomes, heads of IT can unlock the true potential of the technology and create a workforce ready for the future.

More on building AI skills:

• IT leaders rethink talent strategies to cope with AI skills crunch
• 10 most in-demand generative AI skills
• 74% of workers suggest employers to blame for their AI skills gap
• 10 generative AI certs and certificate programs to grow your skills
• The real AI training gap? IT leaders believe in it, but many don’t provide it
  

>