Hundreds of millions of workers use AI tools such as ChatGPT and Gemini every week. That’s the good news. The bad news is that most are wasting the opportunity.
According to The AI Proficiency Report by Section, only 3% of AI users qualify as practitioners and less than 1% are experts capable of building reusable prompts and automated workflows. The rest are novices and experimenters using large language models (LLMs) as a glorified search bar and grammar checker.
This AI proficiency gap is creating a return on investment (ROI) ceiling for organizations. When 97% of the workforce is only scratching the surface, enterprises cannot realize the AI transformation they envisioned.
Until teams move past basic queries and into sophisticated, automated workflows, the true potential of AI will remain a line item on the budget rather than a needle-mover for your business.
According to the Section report, there are three main reasons for the big gap between having an AI tool and getting value from it:
- 88% haven’t received AI training
- 86% can’t officially access AI at work
- 82% say their managers discourage AI use
Adding to the friction, many organizations have not, or cannot, identify a strong use case, according to a Harvard Business Review (HBR) article. Efficiency gains from deploying AI for tasks such as summarizing documents, finding information, and curating emails are important but are the floor, not the ceiling.
These incremental gains are easy to measure and justify, but they’re exactly why so many AI programs are stalling. Saving 20 minutes summarizing emails is worthwhile, but it's not the same as asking why that email chain exists in the first place or whether the underlying process could be automated entirely. It’s the difference between using AI and understanding how AI is used.
A February HBR report summed it up: “Employees experiment with new tools but don’t integrate them deeply into how work actually gets done, leaving executives increasingly concerned about ROI.”
Practitioners and experts don’t open an AI tool and randomly start typing. They start with a problem. What decision am I trying to make? What does a useful output look like? What context does AI need to get me there? That framing—treating AI as a thought partner rather than a search bar—is the difference between a tool that saves time and one that changes how you work.
In practice, practitioners build. They create reusable prompts for recurring tasks, chain AI actions together into automated workflows, and document what works so that other teams don't have to start from scratch. Now imagine what that looks like at scale.
Picture two organizations of the same size, in the same industry, with the same AI tools and roughly the same budget. In company A, AI proficiency looks like it does in most companies today: A handful of rock stars are doing impressive things in isolation while the majority of the workforce uses AI to clean up emails and summarize meeting notes. The ceiling is low because the floor is crowded.
In company B, half the workforce has crossed the practitioner threshold. The marketing team uses AI to both draft copy and build repeatable workflows that accelerate the launch of a campaign. In HR, onboarding is an automated sequence that:
- Routes tasks to IT, finance, and facilities simultaneously
- Flags missing steps in real time
- Sets up new hires to be productive on day 1
In finance, analysts aren't manually pulling budget data from three different systems. They've built AI agents connected to a unified enterprise data layer that surface answers in seconds and trigger next actions automatically.
The difference between company A and company B isn't the technology. They’re both using the same technology. But company B has embedded AI into how work gets done.
Here's where it gets interesting from a business standpoint. The ROI gap between these two organizations isn't linear. It doesn't look like one company is 50% more productive than the other.
When half your workforce is operating at a practitioner level, the gains compound like interest. Faster workflows free time for higher-value work. Reusable prompts and automated processes become shared enterprise assets—living inside the platform where work actually happens—that can improve and scale.
AI experts are obviously better than novices at writing prompts. But more than that, they’re fundamentally changing how they work. Here are some ways AI rock stars are revolutionizing their work:
- They think in workflows, not stand-alone tasks. Instead of asking for a single email, experts build chains. For example, "Analyze these 10 customer feedback notes, categorize by sentiment, and draft three distinct response templates based on our brand voice.”
- They analyze from multiple perspectives. Power users might ask AI to critique a plan from the perspective of a skeptical chief financial officer or to identify the three biggest blind spots in a strategy.
- They think counterintuitively. They ask AI to generate the worst possible version of a project first. By seeing what to avoid (jargon, clichés, bad logic), they can guide AI toward a superior end result.
- They build reusable assets. Experts develop custom instructions and prompt libraries that can be reused across the enterprise.
Proficiency doesn't happen organically. If 88% of employees haven't received AI training and 82% say their managers actively discourage AI use, the path forward should be a deliberate, structured path from novice to practitioner and even expert—and making that investment an everyday priority, not an afterthought.
Here's the harder truth: Most organizations are trying to build AI proficiency on top of workflows that were never designed for it. You can train employees to write better prompts, but if the data they need lives in three different systems, if approvals still happen linearly over email, and if there's no visibility into how work actually moves through the organization, the ceiling stays low.
This is the enterprise AI challenge hiding in plain sight. It’s not a talent problem. It’s a systems problem. When AI has nothing solid to work with, neither do the people using it.
This is where the ServiceNow AI Platform can change the equation. ServiceNow doesn’t just layer AI on top of existing processes; it’s built the infrastructure underneath them, giving AI something solid to stand on before it's asked to make decisions that matter.
By connecting data across the organization and enabling end-to-end visibility, the platform creates an agentic layer that orchestrates tasks across systems. This means AI isn’t just generating text; it’s getting things done.
That’s what moves companies from experimenting with AI to actually building with it. Proficiency is a training problem and an infrastructure problem. The organizations that solve both are the ones that will stop talking about AI ROI and start reporting it.
So, how do you move the needle and go from 3% practitioners to 70% or even more? It starts with giving your organization the governance, training, and tool access that most are currently missing. Here's a practical blueprint, drawn from Section's research:
First, your organization needs a clear set of AI rules, including how people should use it and for what. Without this, AI use goes underground, which creates risk and makes it nearly impossible to scale.
- Define your "why" and identify the specific business problems you need AI to solve.
- Establish governance: Write a formal AI usage policy and name an executive sponsor who owns the outcomes.
- Diagnose the gap: Audit your current proficiency levels. You can't fix what you haven't measured.
- Provide universal access: Give every employee access to enterprise-grade, approved AI tools. You can't build proficiency with tools people aren't allowed to use.
Once the tools are in place, the goal is to move employees from simple tasks such as writing emails and summarizing docs to genuine problem-solving. This is where proficiency actually develops.
- Roll out targeted coaching: Offer role-specific coaching built around the work each team does.
- Certify your practitioners: Create an internal certification program that gives employees a recognized credential for AI proficiency, and a reason to pursue it.
- Build a use case library and a centralized, searchable library of workflows that the whole organization can use and build on.
- Gamify success: Make it fun. Launch an "AI champions" program to recognize your rock stars and pair them with teams that are still finding their footing.
This is where the real returns show up—not from incremental efficiency gains, but from redesigning how work flows through your entire organization.
- Map your opportunities by auditing your highest-friction processes—onboarding, procurement, data reconciliation, approvals—and ranking them by impact and feasibility.
- Integrate your AI into a unified, secure automation platform such as the ServiceNow AI Platform. This is your AI control tower: a single layer where agents operate across systems, decisions get audited, guardrails get enforced, and humans stay in the loop without becoming the bottleneck. Isolated AI generates text. An AI control tower makes it execute.
- Turn your practitioners into builders and measure what moves the business. Your most experienced operators know where the real friction lives. Invest in turning them into low-code workflow builders so that domain expertise becomes deployable.
- Hold your automation program to the right standard: Time saved gets you a pilot, but speed to market, error rate reduction, and output volume growth get you a budget. Those are the metrics that earn executive sponsorship and sustain investment at scale.
The HBR research puts it plainly: What separates organizations that succeed with AI isn't better tools; it's a more honest understanding of how people experience and adopt them.
We've been here before. When cloud and software as a service (SaaS) first arrived, the early adopters who saw the biggest returns weren't the ones who used them to do the same things cheaper. They were the ones who rewired their processes around them. The same applies to AI.
Put real governance in place. Invest in training. Connect your systems. Give your teams a secure enterprise platform. When you do that, you stop asking why AI isn't working and start seeing it produce results.
Here's the real opportunity: If most of your workforce is at the novice level today, you haven't lost ground. You just haven't started yet.
Find out how ServiceNow can help you put AI to work for people.
