Employees are asking AI first — what that really means for internal tooling

A survey making the rounds this month is being summarised as AI overtaking managers as the first place employees go for advice. The underlying numbers are more interesting than that headline, and considerably more actionable if you are responsible for internal tools.

What the data actually says

The research, conducted by YouGov across 1,068 Australian employees between January and June 2026, found that 44% of employees say their choice between AI and a human depends on the task, while 42% still default to a manager or colleague first. Among senior leaders and business owners, the willingness to ask AI first roughly doubles, to 26%.

So AI has not overtaken managers. A plurality of workers have become task-routers: they decide, case by case, whether a machine or a person will answer better. That is a meaningfully different behaviour, and it is the one worth designing around.

Two caveats before drawing conclusions. The sample is Australian, so applying it directly to other labour markets is a stretch. And self-reported preference is not observed behaviour — people are unreliable narrators of their own habits.

The reasons matter more than the ratio

Employees cited speed (57%), convenience (44%) and better-quality answers (44%).

None of those are statements about artificial intelligence. They are statements about the alternative. If asking a colleague means waiting four hours, finding the right person, or digging through a wiki that was last accurate in 2023, then a tool that answers instantly wins on availability before quality is even considered.

Read that way, the finding is a fairly damning review of internal knowledge systems. The competitor to your intranet is not another intranet. It is a chat box that always replies.

The risk this creates

The obvious hazard is confident wrong answers on policy questions — leave entitlements, expenses, escalation procedures — where a plausible-sounding fabrication does real damage.

The less obvious hazard is drift in what "the policy" even is. When a general-purpose assistant with no access to your actual documents answers an HR question, it answers from the internet's average company, not yours. Do that a few thousand times across an organisation and you have an informal policy layer nobody wrote, reviewed or approved.

And because these conversations are private by default, the usual signal that something is wrong — someone asking a manager an odd question — never appears.

What to do about it

Ground the assistant in your own documents. The gap between a model answering from general knowledge and one answering from your actual handbook is the entire difference between useful and dangerous. Retrieval over your own sources is the minimum bar.

Fix the source before automating access to it. An assistant grounded in a policy library that is three years stale will confidently deliver stale policy, faster and at greater scale. This is the step most often skipped.

Instrument the questions, not the answers. The aggregate log of what people ask is one of the better organisational diagnostics available. Repeated questions on one policy mean the policy or its communication is broken.

Be explicit about the boundary. Say plainly which questions the assistant is authoritative on and which require a human — and make escalation one click, not a dead end.

Do not treat this as a manager problem. Nothing here suggests employees value managers less. It suggests they will not wait four hours for a factual answer that could take four seconds. Those are different issues with different fixes.

The buying implication

If you are evaluating internal knowledge or HR-assistant tooling, the differentiator is not model quality — most vendors sit on similar frontier models. It is grounding: what the tool can see, how permissions are respected, whether answers cite their source, and what happens when it does not know.

A tool that says "I don't have that, here is who does" is worth considerably more than one that always produces an answer. The survey shows employees will use whatever answers fastest. That makes it your responsibility to ensure the fastest answer is also the correct one.

---

Source: Human Resources Director on the YouGov findings