Here’s what we believe about good thinking.
Set aside your assumptions. Look at the facts objectively. Let the evidence speak for itself. Reach conclusions without bias. This is the ideal, the view from nowhere. The neutral observer who sees reality as it actually is, uncontaminated by prior interpretation.
It’s also impossible.
Structurally impossible. Because before you can observe anything, you are already inside a framework that determines what counts as a fact, what counts as evidence, what the relevant features of the situation even are.
You don’t see and then interpret. You see as. Every perception arrives already interpreted. Already shaped by the background, the thrownness, the history of structural coherence we’ve been building across the last five videos.
Gadamer called this the hermeneutic circle. You can only understand a part by reference to the whole, and you can only understand the whole by reference to its parts. There is no entry point outside the circle. No view from nowhere. Interpretation is not something you add to neutral perception. Interpretation is perception.
This is where Winograd and Flores make their sharpest break from mainstream computer science.
The entire project of building intelligent systems assumed you could start with a neutral representation of the relevant domain, objects, properties, relationships, and then reason correctly inside that representation. The representation was supposed to be the objective starting point. The view from nowhere, formalized.
But who chose the objects? Who decided which properties matter? Who determined which relationships are relevant? Those choices weren’t neutral. They were interpretations. Made from a background. Standing somewhere. Already inside a hermeneutic circle that the formal system then sealed shut.
And here’s what sealed shut means practically, once you’ve formalized a domain, the system can only see what the formalization allows it to see. Novel situations that require reinterpreting the base categories are invisible to it. Because the interpretive framework that would recognize them as relevant was never encoded, because it lived in the background of the person who built the system, and backgrounds don’t transfer into formalizations.
This is Gadamer doing real work in software design. The system cannot question its own starting assumptions because questioning assumptions requires standing somewhere outside them, and the system has nowhere outside.
Humans can do this. Partially, imperfectly, with effort. Because we have the background. Because breakdown forces us out of our current interpretive frame and into a new one. Because thrownness means we’re always already carrying multiple frameworks in tension with each other.
That tension is not a bug. It’s the mechanism by which interpretation updates.
When you prompt an AI, you’re operating inside this completely.
You frame the question from your interpretive background. The AI responds from patterns trained on millions of other interpretive backgrounds. Something useful emerges in the overlap. And when it works, it works well enough that it feels like shared understanding.
But watch what happens at the edges. Ask something genuinely novel, not novel in topic but novel in structure. A question that requires the AI to step outside its interpretive frame and recognize that its categories are wrong for this situation. That the objects it’s working with aren’t the right objects. That the whole formalization needs to shift.
It can’t do that. It will give you a fluent, confident, well-structured answer from inside the wrong frame. And it won’t know it’s in the wrong frame because knowing you’re in the wrong frame requires standing somewhere outside it.
This is not a capability problem that more training solves. It’s a structural condition of any system built on formalized representations.
You are not neutral either. But you have a breakdown. You have thrownness. You have the background that occasionally, uncomfortably, forces you to see that your frame was wrong.
That discomfort, that productive wrongness, is one of the things that makes you irreplaceable.
@ Lee Greene