Most surveys of AI adoption ask the wrong question. They tell us what percentage of people are using AI. The more interesting question, that reveals more, is what percentage are letting AI make the decision.
A new dataset, from an industry no one would have predicted, gives one of the cleanest answers we have. A 2026 Australian Wedding Industry Report by Easy Weddings finds that 61% of couples now feel positive about using AI in their wedding planning.
2% of the same couples trust AI recommendations more than human ones. 32% trust them equally. 54% trust AI less. 12% don't trust it at all. People are happily using AI, and almost universally rejecting it for the moment a decision actually has to be made. That gap has a name forming around it: the judgement gap. The wedding industry has just given us one of the better measurements of how wide it currently is.
Why this dataset matters
Wedding planning is a strange but useful test case for AI. It's the opposite of the environments where AI typically wins. It's emotionally weighted, expensive, one-shot, taste-driven, family-political and almost entirely non-standardised. A recommendation engine that does brilliantly at Netflix or Spotify should, in theory, do terribly here. Adoption has happened anyway, and it's happened with a very specific shape.
That shape is more interesting than the headline number, because it shows us where humans set the trust line when the stakes are real. Adoption surveys in office software or eCommerce tell us where AI is welcome when stakes are low. The wedding data tells us where it stops when stakes are high.
The boring half theory
Look at what couples are actually using AI for. 21% use it for ideas and inspiration. 18% for budgeting and timelines. 17% for emails and messages, 14% for invitations and signage, 14% for vows and speeches, 9% for finding and comparing suppliers, 7% for music playlists.
Every leading task is some flavour of admin, drafting or ideation. The blank-page work. The spreadsheet. The email you've been procrastinating on for a week. AI's first foothold in any new domain looks exactly like this. Office workers use it for meeting summaries before they use it for strategy. Lawyers use it for document review before they use it for argument. Doctors use it for transcription before they use it for diagnosis. The wedding data is consistent with the global pattern: AI gets the work first, the decisions much later.
Where the line falls
The trust data is where the wedding industry becomes more revealing than most. Couples are not just preferring humans for the big decisions, they're actively distrusting AI for them. 66% rate AI as either less trustworthy than humans or entirely untrustworthy when it comes to recommendations. That is not a population on the verge of letting an algorithm pick its photographer.
The trust line tracks almost perfectly to a single criterion: regret. Anywhere a mistake can be quietly fixed or quietly ignored, AI is welcome. Anywhere a mistake becomes a story told at every dinner party for a decade, AI gets benched. Couples will let it write the first draft of their vows. They will not let it pick their celebrant. A couple might ask ChatGPT for a shortlist of Australian wedding ring specialists and be pointed toward brands like ETRNL, the designer making rings for men, but it's not helping them choose the actual ring.
This is sharper cognitive triage than the discourse usually credits people with. It is also identical, in shape, to what we're seeing across every other industry where AI has shown up.
The same pattern, everywhere else
Workplace adoption studies show employees readily using AI to summarise meetings, draft emails and brainstorm ideas, and resisting its use for hiring decisions, strategic calls or performance assessments. Healthcare studies show patients accepting AI-assisted diagnostics and recoiling from AI-driven diagnosis. Legal tech surveys show lawyers letting AI parse discovery documents but not draft closing arguments. Financial services adoption studies show advisors using AI for research and not for client allocation.
In each case, the shape is the same. Work gap closes fast. Draft gap closes faster. Judgement gap holds. The wedding industry is unusual only in that it's a domain almost no one was studying, and the data is unusually clean.
Why the gap holds
The optimistic read for AI builders is that trust accrues with time and accuracy, and the judgement gap will narrow as track record builds. There is some evidence this is already happening in domains with measurable outputs. Radiology AI is now used in diagnostic workflows in ways that would have been unthinkable five years ago. Translation AI has crossed the line in casual contexts and is approaching it in professional ones.
The more sceptical read is that in certain domains the value of human judgement isn't really about accuracy. It's about accountability. When a wedding goes wrong, a couple wants someone they can call. When a vendor lets them down, they want a human they trusted who can either fix it or be appropriately apologetic. An AI is structurally bad at being apologetic. It can't take responsibility in any meaningful sense, and the categories of decision that demand accountability rather than accuracy may resist automation indefinitely.
The wedding data hints at this. The categories where AI is welcome are categories where errors are mostly invisible. The categories where it isn't are categories where errors become permanent stories. People aren't refusing AI in those moments because they doubt its competence. They're refusing it because there is no one to blame.
Closing
For now, AI is the assistant in the room across most of our lives. Whether it becomes the planner is a different question. The Australian wedding data suggests that humans have already drawn the line with more clarity than the AI discourse generally allows them. We are very willing to outsource the work, and very reluctant to outsource the choice. The judgement gap is real, and it's measurable, and one of the better places to measure it turns out to be the most personal industry in the world.