There is a reading of the current data that says buyers have embraced AI. There is another that says they distrust it. Both are supported by good research, which usually means the question is wrong.

Buyers are not confused and they are not divided into camps. They are using AI heavily for one part of a purchase and refusing it for another, and the boundary between those parts is consistent enough to plan around.

For anyone selling sponsorship, exhibition space or event services, that boundary is the useful thing in this data.

Usage is no longer the interesting number

MarTech reported in May 2026, citing Exploding Topics data, that 77.6% of consumers have used AI to help with shopping in the past six months, with more than 43% doing so weekly. About 68.64% of those users say it has influenced a purchase they would not have made otherwise.

Those figures put AI inside the shopping habits of most consumers. Whatever debate existed about whether buyers would accept AI in the process has been settled by the buyers themselves.

So the question worth asking is no longer whether they use it. It is what they use it for, and what they refuse to use it for.

The same buyers say the word puts them off

TechCrunch reported in June 2026, on a WordPress VIP survey, that 60% of US consumers find brand messaging containing the term "AI" to be a turnoff. In the same survey, 86% said they do not fully trust AI and still want to explore original sources.

Set that beside the 77.6%. The two figures come from different surveys, so they are not the same respondents. But both describe consumers, and together they describe a market that uses AI constantly and does not want brands advertising that they use it.

That is easy to read as hypocrisy. It is not. It is a distinction between a tool they chose and a tool being applied to them. Buyers are comfortable pointing AI at a market. They are considerably less comfortable being on the receiving end of it, and they are explicit that they want to check the source themselves.

The asymmetry has a logic. When a buyer uses AI, they control the question, they can ask it again differently, and they can discard an answer that looks wrong. When a brand uses AI on them, they control none of that. The same technology is a lens in one direction and a filter in the other, and nobody enjoys being on the far side of a filter they cannot inspect.

B2B makes the division explicit

The clearest version appears in B2B, where MarTech reported in May 2026, on Gartner research, that 70% of B2B buyers prefer a digital, self-service buying experience and nearly half use generative AI tools to research vendors and products. The same research found that 69% rely on sales reps to validate what they found, and that human reps continue to outperform AI in key parts of the buying process.

This is the pattern stated plainly. AI to find and narrow. A person to confirm.

It also explains why both of the earlier findings are true at once. Heavy AI use in the discovery phase is entirely compatible with wanting a human before committing, and with being annoyed by a brand that leads with its automation.

What this means for a sponsorship conversation

If a prospective sponsor is doing what the research describes, most of their evaluation happened before you knew the opportunity existed.

They asked an answer engine which shows serve their category. They compared audience figures. They formed a rough shortlist. By the time a conversation starts, your job is not to introduce yourself. It is to confirm or correct a picture that already exists.

Two things follow. The first is that being findable by those engines is now a commercial requirement rather than a marketing preference, which is the subject of why event websites are invisible to AI search. The second is that the shortlist is usually settled early, covered in the day-one shortlist problem.

The verification step is where a person still wins, and it is the part most teams under-resource because it feels like the least scalable.

The verification step is the product

If buyers are using AI to narrow and humans to confirm, then the confirming conversation is not overhead attached to the sale. It is the part of the sale that AI has made more valuable, not less.

That reframes what a good first call is for. Not a pitch, because they have already read the pitch in aggregated form. Something closer to correcting the record: here is what the AI got right about our audience, here is what it got wrong, here is the number it could not have known.

Teams that treat that call as a formality are competing on the half of the process that has been commoditised, and skipping the half that has not.

It also changes who should take the call. When the meeting was an introduction, the right person was whoever could present well. When the meeting is a correction, the right person is whoever actually knows the audience data well enough to say where the public figures are misleading. Those are not always the same person, and on most teams the second one does not sit in sales.

The exhibitor version of the same pattern

Sponsors are not the only buyers running this play. Exhibitors do it too, and the stakes for an organiser are arguably higher because the decision repeats annually.

An exhibitor deciding whether to return is running a version of the same process. They will ask an assistant what the show's audience looks like, compare it against alternatives in their category, and form a view before anyone from the sales team makes contact. What they cannot get from a model is whether the specific people they need were actually in the room last year.

That is the number a human can supply and a model cannot, and it is usually sitting in the organiser's own data rather than anywhere public. Teams that can answer "here are the twelve companies in your target segment who attended, by name" are operating in the verification layer. Teams that answer with total attendance are competing in the discovery layer, against a machine that already has the total attendance figure.

What the research does not settle

It is worth being precise about the limits of these numbers, because they are consumer findings being read across to a B2B context.

The 77.6% and the 60% both describe consumers, though they come from two different surveys. The B2B self-service and validation figures come from a separate study of business buyers. The pattern across them is consistent, which is why it is reasonable to treat them as describing the same underlying behaviour, but they are not the same population and should not be quoted as though they were.

What none of this research measures is how the split behaves for a purchase the size of a sponsorship. A consumer buying a jacket and a marketing director committing six figures to a pavilion are running very different risk calculations, and it would be surprising if the verification instinct were weaker at the larger number rather than stronger. That is a reasonable inference and it is not a measured finding, so it is worth holding loosely.

Disclosure and the messaging problem are separate

It would be easy to conclude from the 60% figure that brands should hide their AI use. That is the wrong conclusion, and the distinction matters.

The complaint is about AI as a selling point, not AI as a disclosed mechanism. Saying "we use AI" as a differentiator is what the 60% are rejecting, because it says nothing about what the buyer gets. Telling someone that the assistant answering their exhibitor question is an assistant is a different act, and the research on that points the other way, as covered in what buyers accept from brand AI.

One is a claim about you. The other is information the buyer needs in order to calibrate how much to trust the answer. Buyers punish the first and reward the second.

The practical test is whether the sentence would still be worth saying if every one of your competitors could say it too. "We use AI" passes that test for nobody. "This reply was drafted by an assistant and checked by our operations lead before it reached you" says something specific about how the organisation works, and it is the kind of statement that survives being copied.

Where this leaves an event team

The practical shape is not complicated, though it cuts against how most teams currently allocate effort.

The discovery layer needs to be machine-readable, because that is where the shortlist forms and no human from your team is present. The verification layer needs to be unmistakably human, because that is what buyers are explicitly holding out for and what they are least willing to accept from a model.

Most teams have this the wrong way round. They invest in polished human-authored marketing that AI crawlers cannot parse, then hand the confirming conversation to a template.

The 77.6% and the 60% are not in conflict. They are telling you which half of your process to automate and which half to protect. That division is the one we build every engagement around, and if you want to work out where the line sits for your show, book a call.