Stop chasing AI experts

If Nike hired Michael Jordan to work at headquarters, would you expect the marketing team to start sinking three-pointers?

Of course not. He’s extraordinary, but skill doesn’t spread by proximity.

Here’s a better question: What do Nike employees need to know about basketball? The rules. Game duration. Equipment specs. Enough to design better shoes, write sharper campaigns, and forecast demand accurately.

They don’t need to play in the NBA. And Nike doesn’t need to hire NBA players to improve its business.

The same is true for AI. Most companies don’t need extreme AI talent to unlock real efficiency gains. They need people across the organization to understand how AI applies to their work.

Until leaders get specific about which AI skills matter, where they live, and how they show up in day-to-day work, no amount of hiring AI experts will make an organization truly AI-enabled.

THREE TYPES OF AI SKILLS

“AI skills” aren’t a single capability. In practice, it’s three categories, each with its own learning curves, and business outcomes.

1. AI literacy: Everyone’s baseline

2. AI integration: Technical professionals’ everyday craft

3. AI creation: Specialists’ deep work

1. AI literacy is everyone’s job

I like to think of this as teaching the entire company how to drive with GPS. Not everyone needs to build the map. But everyone should know when the directions are reliable, when the route is risky, and when the system is confidently wrong.

First is AI literacy. Employees need to understand what AI can do, what it can’t, and what it will do when it doesn’t know the answer. Literacy prevents common failures: over-trusting outputs, under-using tools, and feeding poor context.

Second is AI tool fluency, which is role-specific. A marketer generating content, a recruiter screening candidates, and a support lead drafting responses all need different AI tools.

One reason I like IKEA’s approach is that they’re treating AI literacy as every employee’s responsibility, and the company’s responsibility to enable it. They equipped thousands of coworkers with Microsoft’s generative AI tools and gave them time to learn.

What did this look like in practice? Designers generate product visualizations, store managers create training presentations, and supply chain analysts draft forecasting reports.

Everyone, not just one department.

2. AI integration is a core skill for technical teams

If AI literacy is “drive with GPS,” AI integration is “install the GPS into the car.”

This is where engineering teams earn their keep. Integration skills include prompt design, system evaluation, and knowing when AI belongs in the flow.

Here is what this looks like when done as a system. Salesforce created an internal demo series called Thoughtluck Thursdays, where engineers show short, practical demos of how they integrated AI into their processes and then share patterns other teams can reuse.

Salesforce’s approach works because it creates repeatable templates and guardrails other engineers can ship.

3. AI creation is a specialty, not a company-wide requirement

AI creation is the ability to develop, train, and refine models. It requires deep expertise in data collection and preparation, model training, evaluation, and specialized techniques.

It is also the smallest cohort of most organizations.

If you are not building models as a core part of your product strategy, you do not need a large AI creation team. You need a small number of specialists, and the rest of the organization needs to become competent in usage and integration.

EXTERNAL HIRING HAS ITS PLACE

Let me be clear: External hiring isn’t wrong. It’s necessary when you need skills you genuinely don’t have, especially in AI creation.

But hiring people with “AI skills” on their résumés cannot be your primary path to AI literacy and integration.

First, there is no established market for AI skills. The capabilities are too new, the demand is everywhere, and the talent pool is impossibly thin. Every company is chasing the same small group of people, and most of those people are already employed or starting their own companies.

Second, it is harder to teach someone the ins and outs of your business than to teach them how to incorporate AI into their daily work. The biggest returns come from reskilling the people who already understand your business, your culture, and your systems.

This is where hiring and training stop being separate motions and start becoming one system.

HR OWNS THIS

Don’t get me wrong, IT teams are essential. They evaluate vendors, manage security, and integrate systems. But selecting the right tools doesn’t determine whether AI changes how work gets done by the people doing the work.

Building the right skills does.

That’s why HR needs a seat at the table from day one to ask the right questions: Who gets trained first? How will we train them? Which roles evolve? How will performance be measured? Are there larger talent mobility needs?

Here’s where to start:

1. Pick one team. Choose a group that’s already eager to experiment, has clearly defined processes already, and can measure impact.

2. Give them three months and a small budget. Let them explore AI tools relevant to their work. Provide training. Remove barriers. Measure what breaks and what works.

3. Share the results company-wide. The wins, the failures, the unexpected friction points. Make it real and specific.

That’s your AI strategy. Not a nine-figure hire or a top-down mandate or a hope that capability spreads. Build the skills where work happens, scale what works, and repeat.

Tigran Sloyan is the CEO and cofounder of CodeSignal.

View the full article