Ambient Information

Most of the time, we’re content to simply know that information exists; we lack the interest and patience to actually know the information itself.

But there’s an ambient information reality distortion field.

A good example of this is the rapidly metastasizing rat king that is the ChatGPT “expertise” space. The best results that come from chatbot interfaces with LLMs — large language models — are not instant. They are the output of highly specific prompts and a lot of trial-and-error. That isn’t to write off their utility, nor to ignore the fact that trial-and-error will lead to faster, repeatable results. But it does create an illusion of the availability of instant results. Remember, the AI will give you what you ask for. If you don’t actually know what you need, don’t expect to get a miracle. Right now, using AI is more like the work of an electrician than it is like flipping a switch.

But there’s a very real satisfaction that comes from knowing that someone has already figured out something you didn’t know and has documented it for you. It’s such a powerful suggestion that it often short-circuits your inquiry. Oh, you think, I don’t need to know this after all. I can just reference this article or video later.

I have done this! We all have.

What’s worse, though, is that many of the so-called experts are banking on it.

They don’t want you to know that their quick results were an illusion — that they left out the actual many, many rounds of prompt and output rejection that were necessary to reaching the result they did show. With ChatGPT in particular, this sort of deception is endemic. But I’m not sure it’s entirely malicious.

When what you see and hear all around you are either warnings of AI-enabled irrelevance or claims of a fast-track to new expertise, the choice is clear. I think those who are over simplifying their own experiences with AI in order to demonstrate “expert-level” use are doing it because they believe in the posturing of people just like them. It’s not that the emperor has no clothes, but they’re certainly not the ones in the pictures. Yes, AI can save time, but it will take more time than you think first.

That’s the deception of ambient information — we think it will just soak in somehow. That because it’s out there, it’s in here, our heads, our hands, or our machines. But you can’t shortcut actual learning.

I’ve been a practicing designer for twenty years now. I came of age with a thriving web, with nearly complete access to all the design information available to those who desired it. And yet, even today, I find myself explaining the basics on a very regular basis to other designers. The Basics, as in the elements of design — line, shape, space, form, tone, texture, pattern, color and composition — and the principles of design — balance, pattern, emphasis, movement, and proportion — and how they can be used to assist communication. It’s school stuff! It could not be more accessible, nor more ambiently present in the “design world.” And yet, it’s not well-known or reliably practiced.

Commit to knowing the things you need and want to know.


25 June 2023