I have a group of friends. We plan trips together. One of us has a personal assistant. None of us think twice about giving our friend’s assistant private details to help with travel. We implicitly trust her.
AI agents can do many of the tasks my friend’s PA is doing. Yet bringing my agent into the group chat feels bad. It would immediately raise questions:
What is this?
Is this conversation private?
How will this information be stored, and who will have access?
The agent in chat would give people the ick.
This leaves us in a strange spot. We clearly want agents to help, but the choices today are not great.
We can do everything by hand, the terror!
We can hire expensive humans.
Or we can let our agents impersonate us.
Having an agent pretend to be us is a violation of trust. Yet many people are already doing this, and it seems fine only because it isn’t explicit. Funny how saying the quiet part out loud (“an agent has entered the chat”) changes something that seems fine into something that feels bad.
The problem is that the cultural norms around agents are in flux. We’re stuck in this weird adolescent stage where the technology works, but we haven’t agreed on the social norms yet.
This isn’t a new problem. Norms lag behind technology.
In the early 2000s, taking a picture of someone without asking was inappropriate. Now everyone has a camera, and people accept that in public, someone’s recording. People adapted.
Society will adapt to agents too.
Once the norms are set, I hope that not having an agent in chat will be seen as weird.
Agents are ready. Let’s let them in!
1 thing we’re seeing from companies to accelerate this acceptance is humanizing the products, giving them little faces like grok bot, the Claude lobster, and sora face.
The more an agent feels like a thing with emotion and sentience the more accepting we will be I think.
I agree let the agents in
Who will set the norms? And what are the norms?