A new Gartner survey has put a number on something business leaders have been circling around for months: 22% of Chief Human Resources Officers say at least one business unit in their company has stopped hiring entry-level staff because AI is doing the work instead.

The finding comes from Gartner’s Q4 2025 survey of 110 HR heads, published July 27, 2026. It is one of the clearest signals yet that AI automation is no longer a hypothetical threat to junior roles — it is a live policy decision happening inside real organizations right now.

What makes this data particularly worth paying attention to is not the 22% figure in isolation. It is what sits alongside it: 95% of those same organizations say they have implemented AI in some capacity, but only one in five has realized significant or transformational value from it. The companies cutting junior headcount are not, on the whole, the ones winning with AI. They are organizations that cut costs in the short term without building the foundations that make AI actually work.

What Is Actually Being Automated

The reason entry-level roles are bearing the brunt of this shift is structural. AI is most effective at handling well-defined, repeatable tasks with clear inputs and outputs — which is exactly what junior roles have traditionally been designed around.

Data entry, invoice processing, first-pass document review, basic customer query handling, routine report generation: these are the workloads that large language models and agent-based systems can now handle with reasonable reliability. When a business automates a workflow that would have previously been staffed by three junior employees, the math is straightforward.

But there is a catch that the headline number obscures. The same Gartner research points out that this creates a mismatch between what organizations now have as entry-level work and the traditional profile of people who fill those roles. If you remove the lower-complexity tasks, you are not left with no work. You are left with higher-complexity work that junior employees were not being developed to do, because the simpler work was the development path.

Junior roles were never just about output. They were about progression. Take away the simpler tasks and the pipeline for building capable mid-level and senior staff starts to narrow.

The ROI Problem Is the Real Story

Here is what should catch every business leader’s attention: only 20% of organizations implementing AI are seeing significant returns.

That gap — wide AI adoption, narrow value realization — is the defining business challenge of this moment. Most organizations are spending on AI tools, cutting costs in specific areas, and calling it transformation. The companies that are actually pulling ahead are doing something different: they are treating AI as a system change, not a replacement for specific people.

The difference shows up in how they approach data. Effective AI deployment requires clean, well-structured, accessible data. It requires people who understand what the models are doing and can direct them effectively. It requires clear governance around when AI should and should not act autonomously. These are not plug-and-play capabilities. They take deliberate investment in data infrastructure and in the skills of the people operating these systems.

What This Means for Business

For any business leader reading the Gartner headline and thinking about their own hiring plans, three things are worth sitting with.

The savings are real, but the risks are underpriced. Cutting junior headcount reduces payroll. It does not reduce the complexity of running your business. If AI agents handle those tasks without adequate human oversight and the infrastructure to catch errors, the failure modes are harder to detect and more expensive to fix than a junior employee making a mistake.

The data skills gap is getting worse, not better. As AI takes over execution of simpler tasks, the remaining work is more analytical, more strategic, more dependent on judgment. The ability to work with data — to understand what it is telling you, to identify when a model’s output is wrong, to frame the right questions — is becoming a core business competency rather than a specialist skill. Organizations that are not actively building this capability in their teams are falling behind.

Early career development needs redesigning, not abandoning. Some organizations are using this transition thoughtfully, redesigning entry-level roles to focus on AI supervision, output review, and higher-value tasks that require human judgment. This is harder to implement than a hiring freeze, but it is the path that builds organizational capability rather than just cuts costs.

The Upskilling Imperative

The Gartner data arrives alongside a broader pattern visible across multiple surveys and earnings calls: the organizations generating real returns from AI are the ones with a strong foundation in data literacy and a workforce that understands how to apply AI tools effectively.

This is where the strategic calculus gets interesting for business leaders. Investing in AI tooling without investing in the people who operate it is increasingly being shown to produce mediocre results. The 80% of organizations that have not realized transformational value are mostly not failing because they chose the wrong software. They are failing because the human side of the equation — data skills, analytical judgment, change management — was not developed in parallel with the technology rollout.

For companies watching what the 20% are doing differently, the pattern is consistent: data literacy programs, structured AI adoption frameworks, and deliberate efforts to build internal capability rather than simply outsource decisions to models.

The companies that will be hiring strong, productive mid-level employees in three years are the ones that kept developing people through this transition — and that means finding a way to provide the development path that entry-level roles used to offer, in a new form that is built for an AI-assisted workplace.

---

Enterprise DNA helps teams build the data and AI skills needed to operate effectively in this environment. Explore EDNA Learn for structured upskilling programs, or talk to an advisor about building an AI-ready data capability across your organisation.