AI doesn’t ask permission. Given an objective, it considers the tools available and chooses the one that closes the gap.
Watching this happen isolates a feature of human cognition we rarely see by itself. Thinking about whether to do something usually blends two questions, can I and may I. AI answers only the first. What’s left, when the second question is set aside, is a plain examination of capability.
Possibility is a belief about what the world will allow, shaped by norms and by the assumed reactions of others. Capability is what a system can actually execute when it reaches for a tool and uses it. The two are different, and the conflation between them is familiar enough to pass unnoticed.
Before LLMs could generate coherent arguments from a prompt, such a thing passed for impossible. Yet the transformer architecture they rely on had been public since 2017, and training data was abundant on the open web. What was missing was the bet that scaling compute against those pieces would produce sophisticated reasoning, and the willingness to spend what such a bet costs. When the bet paid off, the label flipped quickly. The universe had not rewritten itself; a group of researchers had committed enormous resources to testing a specific hypothesis, and the hypothesis turned out to be correct.
That is what AI makes visible. Building the system required the people doing the work to push past the impossible label. The system itself carries no such belief; it executes against the objective in front of it. Those prior limits were limits of what was believed rather than limits of what was capable, and the difference was invisible until a system without the beliefs started working.
A new capability moves the line between what people call possible and what they call impossible. The real line was always somewhere else: at the boundary of what could actually be done, which tends to be further out than belief admits. When the line adjusts toward its real location, people feel the ground shift, because they had been living inside a smaller circle than the one their capabilities actually allowed.
People who had seen the capability all along, and been told it was out of reach, welcome the shift. People whose livelihood or expertise was built around the old position of the line resent it. The line moves at the speed of capability either way.
Capability draws the line between possible and impossible. The remaining choice is personal, whether to live inside the limits of belief or the limits of capability.
— Chiaroscuro Joven