AI use in events is growing. Why do professionals feel less capable?

Your team uses AI to draft messages, summarise meetings or prepare proposals. It handles more tasks with it every month. But when you consider letting it answer an attendee without supervision, doubts appear.

That gap between using a tool and feeling capable of applying it also appears in industry data.

More use, lower self-rated ability

In Meetings Outlook Q1 2026, MPI compares several waves of its survey:

Indicator	Q1 2025	Q1 2026
Regular use of generative AI	43%	70%
Own ability rated good or excellent	43%	33%
Overall positive attitude towards AI	68%	62%

Regular use rises by 27 percentage points. Positive self-rated ability falls by 10 points. A favourable attitude towards the technology also declines, although it remains the majority view. These figures appear on page 6 of the original MPI report. You can explore the source in Athena.

The Q1 2026 wave contains 179 responses: 54% planners and 46% suppliers. This is a comparison between survey waves, not a study tracking the same people. It does not represent every events professional.

Are professionals losing trust in AI?

The survey separates two things we should keep distinct: attitudes towards AI and professionals’ assessment of their own ability to use it.

The 33% does not measure how many people consider an AI-generated answer reliable. It measures how many rate their own ability as good or excellent. Saying that “trust in AI fell from 43% to 33%” would therefore change the meaning of the finding.

The authors suggest that the variety and depth of available tools may feel overwhelming and that skills might not be keeping pace. That is their interpretation. The comparison does not establish why each person changed their assessment.

At Sinapser, we see a possible explanation: the more you use AI, the more situations you encounter where judgement matters when deciding what to delegate. That is a hypothesis, not a survey conclusion.

Drafting a message and answering an attendee require different controls

Imagine AI proposes an invitation for your attendees. You can read, correct and approve it before sending it.

Now imagine it answers someone asking whether their ticket includes access to a private dinner. It needs current information, that person’s permissions and a way to recognise when the team should handle the answer.

Both uses can save work. The second also requires you to define what information it may consult, what actions it may take and how a person intervenes if something goes wrong.

Our reading of the study is that increased use alone does not tell you whether a team is ready to take on that responsibility. Adoption figures describe part of the picture.

A useful question for your team

Instead of only asking “Do we use AI?”, try a specific situation: what would need to be in place before we let it answer our attendees directly?

The answer may reveal that you need a clearer information source, access limits for each user or a process for questions the agent cannot answer. It may also confirm that reviewing every output remains the best approach for a particular task.

Caution does not have to mean rejection. It can help you identify the conditions you need to use the technology.

The next article brings that question into daily event operations: what can go wrong when AI answers your attendees, and what controls should you require?