Executives Are Playing With AI, Not Handing It the Big Decisions

On paper, it looks like AI is marching steadily into the C‑suite. Strategy decks are full of AI slides. Boards are asking hard questions about automation and transformation.

But when you look at how leaders are really using AI, a different story emerges: business leaders are more than happy to let AI tidy their inbox—but they are not delegating the core decisions that define their business.

That is the central signal from “Business Leaders Still Primarily Use AI for Simple Tasks,” a new survey conducted by General Assembly and EZRA, two LHH (HR/talent agency) brands focused on practical skills training and coaching. The study surveyed 524 business leaders (directors, vice presidents, and C‑suite executives) in the U.S. and U.K., all from organizations with at least 100 employees, on April 8, 2026.

What the Study Actually Measured

The survey asked leaders how they use AI today, how confident they feel in their AI capabilities, and whether they have received leadership‑specific AI training. It builds on similar surveys of this audience conducted by General Assembly in 2024 and 2025, allowing them to track shifts in behavior and sentiment over time.

The topline finding is clear: AI is common in leaders’ workflows, but mostly for simple, low‑stakes tasks. The three most frequent uses reported were:

• Searching for information (69%)
• Summarizing documents (68%)
• Drafting emails (58%)

By contrast, strategic and analytically demanding applications are much less common:

• Scenario planning (27%)
• Organizational design (27%)
• Financial modeling (28%)

So while AI is present in leadership work, it is not yet embedded in the activities where leaders make the calls that carry real risk, trade‑offs, and accountability.

The Line Leaders Won’t Cross: Judgment vs. Chores

Seen through a human‑behavior lens, the pattern is striking. Leaders are comfortable handing AI their words, but not their judgment. They will let AI:

• Clean up a board memo
• Condense a 40‑page report
• Pull background information before a meeting

But when the task shifts from “help me process” to “help me decide,” leaders largely pull AI back. Strategic choices about structure, scenarios, and capital allocations remain mostly human territory in this sample.

This matters because it contradicts a common fear narrative: the idea that executives are about to offload decision‑making to opaque models. In this study, they are doing the opposite—they are using AI as a thinking aid, not as a deciding agent.

What Changes When Leaders Get AI Training

The study also draws a sharp contrast between leaders who have and have not received leadership‑specific AI training. Among those who have completed such training:

• 96% say their teams use AI tools regularly, versus 82% overall.
• 88% say they know how to use AI without compromising company data, versus 68% overall.
• 97% feel equipped to make AI‑related vendor decisions, versus 83% overall.
• 71% say they have a clear rubric for what “good AI usage” looks like, versus 52% overall.
• 89% feel confident assessing employees’ AI competence, versus 71% overall.

In other words, training does not push leaders to abdicate decisions—it makes them more deliberate about how AI is used, where it is trusted, and how it is governed. Trained leaders also report more use of AI for nuanced tasks like qualitative data analysis and “vibe‑coding” sentiment, suggesting they are experimenting more at the edges of judgment while still retaining final accountability.

Leaders without leadership‑specific AI training, by contrast, are even more skewed toward basic use cases. They are more likely to use AI primarily for drafting emails (75%), and less likely to use it for complex tasks like qualitative analysis (31%) or competitive and market intelligence (19%). Only 19% of this group say they have successfully “vibe coded” something, compared with 49% among trained leaders.

Why This Matters for the Human Edge

For organizations concerned with the human edge in an AI‑saturated environment, this study sends a nuanced message. On one hand, executives are clearly not surrendering decision‑making to AI; they are drawing an intuitive line between what can be automated (search, summarization, draft text) and what must remain human (judgment, trade‑offs, accountability).

On the other hand, the data shows that without deliberate, leadership‑specific development, AI risks being trapped at the level of administrative support. The potential of AI to broaden leaders’ field of view—by surfacing patterns, testing scenarios, or deepening qualitative insight—remains underused.

The implication is not that AI will replace leaders, but that leaders now need a new literacy:

• Knowing when and how to invite AI into the decision process
• Being explicit about the human decisions AI will never make
• Building guardrails, training, and culture so teams treat AI as a partner in thinking, not an invisible decision‑maker

This study from General Assembly and EZRA captures leaders at an important inflection point: AI is in the room, but humans are still at the head of the table. The next chapter will be written by how intentionally organizations develop leaders to keep it that way—while still unlocking AI’s full, strategic value.