Spend enough time on AI X and it becomes easy to forget how much you’ve already learnt.
Agents, context, models, tools. Ideas that once needed explaining become the vocabulary of an ordinary day. You see what other people are building, try things yourself, and gradually change your expectations of what a computer can do.
Then you go to work and discover that concepts you now consider commonplace still need an explanation.
I keep coming back to that gap. Those of us immersed in AI need to be more aware of the small world we inhabit, and more thoughtful about how we bring others into it.
Familiar to whom?
I love the possibilities agents have opened up. Being able to discuss an idea and start turning it into a tool changes your relationship with technology. You begin asking what you actually want, because making something personal starts to feel within reach.
That was the joy behind my last piece about building the computer I wanted. A discussion that might once have ended with “someone should make that” can now become the beginning of making it yourself.
But even that sentence assumes quite a lot.
It assumes you know what an agent is, have some sense of what it can do, and feel comfortable enough to experiment. It assumes you can recognise a useful result, notice when something has gone wrong, and decide what to try next.
When those habits become familiar, they can disappear from our explanation. We describe the possibility and skip the knowledge that makes it possible for us.
At work, I still see people struggling with ideas and concepts I now take for granted. That should make me examine my assumptions. My familiarity tells me something about the time I’ve spent in this world. It tells me very little about where somebody else is starting.
We’ve forgotten how much we’ve already learnt.
The size of the room
There is a useful reality check in the numbers. In Pew Research Center’s survey of 5,119 US adults, conducted in February 2026, 51% said they never use AI chatbots such as ChatGPT, Gemini or Copilot.[1]
That is a specific measure of chatbot use, rather than every encounter with AI. Even so, it is worth sitting with when your feed makes these tools feel like part of everyone’s daily life.
A timeline full of people comparing models and building with agents is a wonderful place to discover possibilities. It is also a very particular room. If we spend enough time there, its conversations can start to feel like a description of the wider world.
Inside that room, we might be debating which tool offers the best way to delegate a task. Outside it, someone may still be trying to understand why they would delegate anything to a chatbot in the first place.
Both conversations matter. The second is easy to overlook when the first occupies most of your attention.
I want to keep the excitement of that room. I also want to remember its size.
Knowing about AI is only a beginning
It is tempting to imagine that people simply need to see what is possible. Show them a sufficiently impressive example and the value will become obvious.
Sometimes an example helps. But adoption asks more of someone than being impressed.
They have to find a use that matters to them. They need time to try it, confidence about what information they can share, and some way of deciding whether the result is any good. A promising demonstration can leave all of those questions unanswered.
Pew’s research reflects that complexity: non-users cited lack of interest, privacy concerns, doubts about accuracy and not knowing how to use chatbots.[1] Those are different obstacles. An explanation of the technology will only address some of them.
There is also a perfectly reasonable question underneath all this: is it worth the effort for what I need to do?
We should be willing to hear the answer. Bringing someone along means giving them enough understanding to make a useful choice. Their choice may differ from ours.
I think enthusiasm becomes more persuasive when it makes room for that.
Bring the context with you
For me, the responsibility starts with those of us who already feel comfortable here.
We can pay more attention to the assumptions hidden in our language. We can explain an unfamiliar concept without treating the explanation as an inconvenient delay. We can remember that what feels obvious now became obvious through exposure and practice.
And when someone does want to try, we can begin with something they care about.
One real task gives an experiment a purpose. Work through it together. Explain the decisions as you go: what context you supplied, what you checked, what you changed, and where you still needed your own judgement.
That makes the process easier to understand and gives someone a better chance of repeating it independently. A finished result can be impressive while leaving its audience unsure how to begin.
I also think we need patience with questions our feeds have moved on from. The fact that we have discussed something hundreds of times does not mean the person in front of us has heard it once.
They should be able to ask a basic question without feeling late to the conversation.
Keep the joy
I don’t want to lose the energy of being around people who are exploring. That energy has helped me imagine and build things I might otherwise have accepted as impossible, or left for somebody else to make.
But I want my enthusiasm to stay connected to the people around me.
If these tools offer the possibilities I believe they do, then helping someone find their own starting point deserves some of the attention we give to finding our next experiment.
Sometimes that will mean slowing down, explaining a term, or helping with an ordinary task. Sometimes it will mean listening to a concern we have become too comfortable dismissing.
The next time something feels obvious, I want to ask myself: what have I learnt that I’m leaving out of the explanation?
[1] Pew Research Center, Why don’t people use chatbots?, 17 June 2026
A father of two, a builder of software, and someone who finds it difficult to leave an interesting idea alone.
Follow along on X ↗
