Articulation of Intent - The Durable Human-Owned Layer
It's clear that we, as humans, are evolving to operate every day in a system that has AI as a part of it. In this system, human + AI, I see it consisting of three main layers in a cycle:
- Articulating Intent
- Understanding Intent
- Connecting Intent to Action
First, why base it around "intent"?
- intent encompasses the human aspects 1 of an idea
- asking someone "what's your intent?" expects an answer that not only explains the what, but also the why
- this is the framing and execution motivation around the idea
- when all else is equal, intent is the differentiator since the framing, why, and packaging of those components are going to be uniquely mine as opposed to yours2
They say history repeats itself, that every quote that can be said and idea that can be had is already taken, but even when an idea has been expressed before, the reason why the individual wants to express it can still be their own.
Second, why is "articulation" the durable human-owned layer?
- technology can improve all three layers, yet articulation remains human-owned because the work is still about discovering and expressing what we want, even when technology helps us do it
- as models and the products around the models improve, they can reduce our effort required to ensure that the system understands and acts on our intent3 -- those are largely the AI parts of the system
- although ai can improve your expression/articulation, you still have to be in tune with your intentions (which change as you try to articulate them!) to judge whether AI's expression carries what you mean
There's a quote in 'Man's Search for Meaning' that goes, "His unique opportunity lies in the way in which he bears his burden." Frankl, the author, is describing how, when suffering is unavoidable, a person can still find meaning through the way they face it. Life is incomprehensibly varied. Therefore a singular opportunity won't be taken upon the same by two different people because we each have our own personally meaningful, unique intentions.
Third, if this were to be true, what should we do about it?
- well, i'm still figuring this part out, but with anything you want to improve there's being aware of it and practicing it, so I imagine at least doing the former is a start
- perhaps this looks like wrestling with your thoughts and feelings around a particular idea longer than you would have in the past, so that you go through all the iterations of grasping your own thought yourself and make future times easier
- or putting more articulations out into the world, seeing the feedback, and then doing it again
- or simply experiencing something, consciously deciding if you like it, then explaining why
- one thing i've learned about effective communication is spending more time thinking about the recipient/other person (i.e. what they care about and what would make them feel a certain way), so your ability to personalize the framing and why for them is likely a key factor in successfully articulating your intent
- also realize that the other layers won't be perfect either, so work to improve those, too, in whatever system you do your work in (but maybe spend less time there, since technology will rapidly improve it for you)
Articulation helps us discover our intent.
Afterword
I think about this quite a bit because I struggle with the ability to effectively speak my thoughts, which is part of articulating my intent. I think I do a better job at meaning what I say, than saying what I mean. In some cases this works out because people can "get what I mean" (they're good "understanders of intent"! - the second layer), but not all. And even still, I'd love to be able to express myself fully, as intended. As I'm sure others would, too.
Footnotes
-
maybe the better word is "soulful", I'm not 100% sure, but when i say this, i think we all have an intuitive sense of what this means. so for now we can describe it as such and visualize the thing that makes our hairs rise and brain tingle. ↩
-
even if technology can take us to the extreme perfect state where there is zero signal loss between our conscious intentions and reality, and those intentions can be perfectly understood and represented in reality, our intentions can still differ. access to the same capabilities doesn't mean we want the same things from them, for the same reasons. ↩
-
this is why I believe people in more ambiguous domains are drawn to Claude models as opposed to OpenAI models because in my experience Claude can understand/infer what we really mean better OOTB. most of the time, we don't really know what we want, so a model that's great at doing exactly what you say (i.e. OpenAI models) often requires more explicit correction/steering to arrive at what you really intended since you don't describe it precisely the first time. ↩