David Graeber's book Bullshit Jobs talks about how our modern economy creates a massive amount of pointless roles - jobs that exist just to tick boxes, "duct-tape" broken processes, or serve as status symbols to make bosses look important.
I don't agree with everything Graeber says. I'm not against capitalism; in fact, I'm an advocate of it. But I am highly anti-theater.
The problem isn't that markets are working too well; it's that dubious incentives create red tape. When we reward headcount, meetings, and slide decks, we get more of them. When incentives are misaligned, people end up playing to metrics rather than solving real problems.
I see this even in my life as a student. I often learn more efficiently by working through lecture material and exercises independently, where I can pause, translate terms, and investigate gaps in my understanding.
I prefer the freedom to choose my own method and be judged on the outcome rather than the ritual of just being present. For me, attendance is useful when the discussion or feedback adds something I cannot get from the materials alone. Other students benefit from the structure of a classroom; the point is to make room for both approaches.
This brings me to AI.
LLMs have lowered the cost of drafting, summarizing, and some routine analysis, although checking the output still takes work. This presents us with a choice:
- Automated Theater: We keep the same useless rituals, but we just do them faster.
- Subtracting the Theater: We delete the low-value tasks and use that free time for real work.
AI won't choose for us - incentives will.
If a role consists mainly of formatting and summarizing, employers may expect software to handle more of that work.
The more useful question is what you can decide, build, or improve with those tools, and whether you can tell when their output is wrong.
Technology can eliminate roles as well as reshape the tasks within them. The danger isn't the technology itself; the danger is keeping the old rituals and calling it productivity.
I expect more roles to involve specifying tasks, checking automated work, and taking responsibility for the result. That makes these qualities more valuable:
- Judgment
- Trust
- Fixing root causes
- Being accountable for results
Competition can punish waste, but misaligned incentives can also sustain it. We shouldn't wait for a recession to question work that adds little value.
We need to create a system where fairness and stability exist alongside this new, leaner productivity. This is where concepts like Universal Basic Income might become necessary. If we are going to automate away the "bullshit jobs," we need a safety net that allows people to retrain and move up the value chain without falling into poverty.
Graeber warned us about meaningless work. AI is a lever. We can pull it the lazy way and just generate faster nonsense; or we can pull it with intent - and free people up for work that actually matters.