Vibe-designing

Large Language Models, aka AI, are the monumental claim that language is everything - that an image is text, a video is text, and somewhere at the bottom of it, even a feeling is text, which is either the most beautiful thing that’s happening to human expression or the worst, depending on how optimistic you are about technology (likely not).

Designers often treat ‘sketching’ as synonymous with ‘thinking’: the hand knows something the brain hasn’t caught up with yet, and that fast loop of making something, iterating, feeling what came back, and responding is a more honest account of where ideas really live. They live BEFORE language.

Think about it, vibe-coding runs on the same logic: AI makes something, you try it, something feels off, you push back; and design directors, for instance, have always steered projects this way too, with a feeling that’s legible to them but hard to transmit. They design with designers as instruments, then sell their vision to the client. Note this: the most articulate ones get to the top, narrative is king.

The interesting question is if the system will be able to feel what your particular off feels like well enough to translate it for you, without the need to articulate. We might be in a pre-graphic-interface era of computers right now, to draw an analogy; everything goes through text, which is ‘efficient’ only to a point.

There’s a good chance that being articulate won’t be necessary in the near future, that’s going to be crazy.