You are going through the deck one last time before the board call. Six recommendations. “According to McKinsey…” “Gartner research shows…” “A recent government report on AI adoption states…”
Reading it back, something is off. There is not a single line that says “our assessment is” or “we have concluded.” The deck is detailed, well-sourced, and thorough. Your own read on what the business should actually do next is missing from it entirely.
That gap has a name. It matters more than the slides do.
What is expertise displacement?
Expertise displacement is what happens when a delegate leads AI recommendations with someone else’s authority rather than their own. The tell is in the language. Every claim opens with a named external source, “According to McKinsey…” or “Gartner reports…” rather than “Our analysis shows…” or “We have identified…”. The delegate’s judgement, if it appears at all, arrives behind the citation.
Expertise displacement sits in the communication layer. A delegate who has done the reading, reviewed the research, and formed a clear view can still fall into the pattern. The habit builds gradually. Early in a new role, when the subject is unfamiliar, leading with named authorities feels like the responsible choice. Over time, if the habit goes unchallenged, it becomes the default mode. Every recommendation arrives pre-endorsed by someone else.
The sources cited may well be correct. What is absent is the delegate’s own view. A board member reading the deck gets a clear picture of what McKinsey thinks about AI adoption in businesses of this size. They come away with less certainty about what the person presenting it thinks.
The delegate who ran the analysis, shaped the recommendations, and has the deepest knowledge of the business’s specific situation is, paradoxically, the least present in their own presentation.
Why does this undercut the authority your role needs?
The delegate role requires the room to believe you hold a view, not merely access to research. When every recommendation leans on external endorsement, you send a signal the room picks up whether you intend it or not. Korn Ferry’s work on AI leadership names this directly. Organisations assign AI mandates to strong operators who lack AI-specific competencies, creating high expectations with limited preparation on both sides.
Spencer Stuart’s research on AI leadership adds another layer. The delegates who build genuine authority are those who treat AI as a subject requiring personal engagement, not one they can absorb through briefings and reports. The gap between those two positions is exactly where expertise displacement takes hold.
The board may not be technically fluent, but it recognises when someone is presenting their own judgement versus relaying a curated set of sources. Both feel different in the room. A recommendation framed as “our analysis shows X and here is the evidence” carries different weight than “the evidence says X and we recommend following it.” The second version puts the board in the position of evaluating the source. The first puts them in the position of evaluating your read.
ESG Dive research found that roughly 61 per cent of senior operators in AI leadership roles fear consequences if they fail to lead adoption credibly. Expertise displacement compounds that exposure. A delegate who has not built visible ownership of their position is more vulnerable when results are slow to arrive, not less.
Where does this pattern actually show up?
The clearest version appears in board presentations and founder conversations, but the pattern runs through everyday communications too. Check the language in your last AI progress update. Who holds the subject position in each sentence? When the expert named at the start of a claim is always external, McKinsey, Gartner, a government report, the delegate’s own assessment is either absent or buried.
In team briefings, the same pattern shows up in how decisions get framed. “Industry analysts say we should prioritise automation in customer service” rather than “we have decided to prioritise automation in customer service because…” In conversations with the founder, it is the choice between “HBR has written about this type of failure” and “my view is that we are about to make the same mistake.”
Listen to the language over the next two weeks. The shift from “according to” to “based on our assessment” is small on the page. The room hears it differently. When a delegate owns the claim, the recommendation carries weight. When the delegate is reporting what they found, the implicit invitation is for someone else to challenge the source, or to find a better one.
The tell is consistent across every communication format. External authority in the subject position, the delegate’s assessment in a subordinate clause, if it shows up at all.
When should you cite and when should you lead?
External research has a legitimate place in any AI recommendation. The question is whether you are using it to confirm a position you already hold, or whether it is standing in for a position you have not yet formed. Authority anchoring reverses the usual order. The assessment comes first, the evidence follows. You own the claim, then bring in the sources to support it.
The structure is straightforward. Before referencing any external report, write your actual view in one or two sentences. If you can state it clearly, the research belongs after it. “We have assessed three options and concluded X is the right approach for this business at this stage. The McKinsey analysis of mid-market AI adoption supports this direction.”
That is a different sentence from “McKinsey’s analysis points toward X, so we are recommending it.” One version presents you as the person with a position. The other presents you as a relay for someone else’s conclusion.
If you reach for a citation before you can state a clear view, that is the signal. The assessment work is not finished. Going back to it is the right move. Presenting the sources as a substitute for the assessment is the wrong one.
A useful test before any AI recommendation: can you state your view in one sentence without citing anyone? “We should prioritise automating client onboarding this quarter because it is our highest-volume, lowest-complexity process and the tools are ready.” If that sentence exists and you have evidence to support it, the recommendation is ready. If the sentence does not exist, the assessment is not.
How does external research fit when you are leading with your own view?
External sources do important work in an AI recommendation when they are in the right position. A Korn Ferry dataset on AI adoption rates tells you something about pace and market precedent. An EY analysis of board-level AI risk highlights what your board is likely to have seen. Both confirm, challenge, or contextualise your own read. They do not generate it.
McKinsey’s research on AI in the workplace covers patterns across hundreds of organisations. Your analysis covers one organisation, in detail, from the inside. Both are useful. The one that belongs in the subject position of your recommendation is yours.
This also applies to how you handle the limits of your technical knowledge. Technical expertise and business judgement are different things. You can be honest about the bounds of your AI engineering knowledge while still owning a clear business position. “I am not a machine learning specialist, but my read of our operational situation is…” is authority anchoring. It leads with your perspective and is honest about the boundary. Reaching for a technical report to substitute for a business position you have not yet formed is a different move entirely.
Credible research strengthens your recommendations. It does its best work when you arrive at the meeting already knowing what you think.
If you want to work through what authority anchoring looks like in your specific context, Book a conversation.
