Our AI show and tell reached the people who already came

We run a regular session where people show how they use AI in their own work. It works, in the sense that people come, and they keep coming.

That is also the problem. When I looked at who was attending, nearly everyone was a returning attendee and few new people were joining. The session was doing a good job for people who were already interested and very little for the people it was meant to reach.

A second series, with a different invitation and a wider audience, drew many people who had never been to the first. The idea was the same. Who it reached was not.

I think the reason is who an open session selects. The people who come are the most active users, and they are not the people the session exists for. The ones who most need it have not yet recognised what AI could do in their own work, which is harder than learning to use the tool, and they are the least likely to turn up to a session about it. Even the people who find the useful cases are not always the most active users. In my experience they emerge once people have access and a reason to try.

It is a lot like church: the sermon reaches the people already in the pews. There is also a second explanation. The people AI could help most are often the most overloaded, and an optional session is the first thing an overloaded person drops. They are not absent because they are uninterested. They are absent because they are busy.

Here is a hypothesis from one observation, not a result. An enablement activity measured by attendance will drift towards the people who need it least, because they are the ones who turn up. A rising headcount can hide it completely. What showed it was counting who was new, not how many came, which is the same question as who belongs in the denominator of an adoption rate. The next move follows from the adoption data: take it to the teams where adoption lags, in smaller sessions for one team or time one to one, and do not just invite them.

A reasonable objection is that regular attendees are not wasted. They spread what they see, and a room of enthusiasts is how a practice gets started. That is true, and it is a reason to keep the session. It is not a reason to report it as reach, because spreading a practice through the converted is a different job from bringing in the people who are not yet there.

The mandate gives a Chief AI Officer adoption through governance and workforce advice as duties. A session that reaches only the enthusiastic is easy to count as progress on both, which is why I now count the newcomers separately.

When you count attendance at your AI sessions, do you know how many of those people were new?