David Graeber’s idea of bullshit jobs has always been an uncomfortable truth. They are roles that exist more to keep up appearances than to create real value. Many jobs are not essential. They persist because bureaucracy, inefficiency, and corporate theatre demand them. Now, with AI, we watch an ironic spectacle unfold: the fear of losing jobs that perhaps never needed to exist in the first place.

Large language models and AI agents are being sold as a revolution, capable of replacing human labour at a scale we have never seen. But if an AI can replace a role, was that role ever truly necessary? If automation wipes out entire layers of corporate work with no real impact, isn’t that proof those roles were digital duct tape over inefficiencies that should never have been there?

There is one thing, though, that AI will not replace: programming. Not in the shallow sense of typing code, but in the deeper act of understanding. Of immersing yourself in a field (medicine, finance, logistics, philosophy) and translating its logic into something computable. Programming is not syntax and automation. It is modelling reality, organising knowledge, making sense of complexity. AI can generate code. It does not think about the problem the way a programmer does.

The real revolution is not AI replacing jobs. It is the realisation that many jobs should never have existed. The real work, the work of structuring thought, organising systems, and creating true value, remains as vital as ever.