AI and the Phenomenon of Bullshit Tasks
William Gibson once remarked that "The future is already here — it's just unevenly distributed."11 Gibson made this remark in a 1999 interview on NPR's Talk of the Nation and has repeated variations of it since. The quote is a bit tired today, having been used by marketers and tech enthusiasts the world over. But the idea that the future isn't some far-off fantasy has its uses. When thinking about what kind of future a technology might eventually bring, we can learn a lot from its initial uses. To temper the wild and, at times, downright embarrassing visions provided by the TechBros, many researchers and journalists are now busy trying to understand the effects of AI as it works its way into different areas of life. When it comes to the many jobs that involve sitting in front of a computer and bashing away at a keyboard — and following this logic that the future is already here — I'm afraid I bring some bad news. Our future involves quite a lot of bullshit. Let me explain.
There is no shortage of speculative thinking around the future AI will bring. In a blog post titled The Gentle Singularity,22 Altman, S. (2025). "The Gentle Singularity." blog.samaltman.com. for example, Sam Altman writes of enormous gains to the quality of life, mostly down to new scientific discoveries and increases in productivity. Indeed, increased productivity is a recurring theme: "the ability for one person to get much more done in 2030 than they could in 2020 will be a striking change" … "We already hear from scientists that they are two or three times more productive than they were before AI", and "If we can do a decade's worth of research in a year, or a month…". Altman does, thankfully, at least gesture toward some "very hard parts" of this super-productive future, "like whole classes of jobs going away", before reassuring his readers these hard parts will be countered by the world somehow "getting so much richer so quickly".
I don't doubt AI is making some people more productive — if by that term we simply mean increasing output. I know programmers who describe real changes to how they go about their work. I've seen the flooding of social media with slop and I've seen the increases in (AI generated) job applications and research grant applications in my academic job. But when I look to how people are actually using AI, in the form of LLM chatbots (Claude, Gemini, ChatGPT, Co-Pilot) I see a more nuanced and rather different picture than the general hyper-productivity imagined by the Altmans of the world.
"people rarely use AI for the important parts of their jobs, but are much more willing when they perceive the task as having little or no value. If someone declares among confidants they 'just used AI', this doubles as an admission that they find the task unbearable."
When it's not forcd upon them by managers, I think people use AI for two main reasons: either for things they want to do but can't (without help), or things they can do themselves but don't want to. Most of the promises around AI and productivity focus on this first scenario, but I think a lot of the more socially interesting stuff is happening with the second: people using AI to do the parts of their job they don't want to do. Most commonly, these are highly bureaucratic or procedural tasks, such as filling in forms, writing reports, preparing formulaic presentations and the like. In my experience, people rarely use AI for the important parts of their jobs, but are much more willing when they perceive the task as having little or no value. If someone declares among confidants they "just used AI", this doubles as an admission that they find the task unbearable. The minimal passable quality that chatbots are able to churn out will suffice, so long as nobody else with eyes on the output cares too much either.
One of the gifts left to us by the late David Graeber was his popular notion of bullshit jobs.33 Graeber, D. (2013). "On the Phenomenon of Bullshit Jobs." Strike! Magazine, 3. Expanded into: Graeber, D. (2018). Bullshit Jobs: A Theory. Allen Lane. For Graeber, a job qualifies as bullshit only if the actual people doing those jobs themselves think the job has little or no social value. A number of studies44 Simon Walo (2023). 'Bullshit' After All? Why People Consider Their Jobs Socially Useless. Work, Employment and Society, 37(5). inspired by Graeber's idea have tried to figure out how many jobs across different societies meet the criteria of bullshit, with one YouGov poll suggesting the number in the UK was 37%. That is a lot of bullshit. But rather than using the criteria of bullshit to classify entire jobs, I think it is even more useful when used as more of a spectrum, from meaningful and important to bullshit, and not only for a job in total but to the range of tasks that comprise any particular job. From bullshit jobs, to bullshit tasks within jobs.
One way AI chatbots are being used are as coping tools for bullshit tasks. We see them popping up wherever there is work nobody wants to do. Not only has AI come to participate in a new hierarchy of work- or task-value, but the use of AI for any particular task should raise the immediate question "Is this task bullshit?" Here, perhaps AI has a really valuable social function, as a detection tool for bullshit. There are already a number of AI detection tools doing interesting things which could easily be repurposed to these ends. Perhaps we could come up with new typologies for bullshit tasks across all jobs, based on where AI is being used in ways that make no contribution to society.
Curiously, towards the end of Sam Altman's vision for the future he arrives at his own 'job theory' — the phenomenon of 'fake jobs': "A subsistence farmer from a thousand years ago would look at what many of us do and say we have fake jobs, and think that we are just playing games to entertain ourselves since we have plenty of food and unimaginable luxuries." It would seem on the surface that Altman has read Graeber, but there are fundamental differences. In Altman's vision, people seem to be intrinsically motivated to do these fake jobs and his future sees them proliferating: "I hope we will look at the jobs a thousand years in the future and think they are very fake jobs, and I have no doubt they will feel incredibly important and satisfying to the people doing them." In Altman's vision, it would seem only the subsistence farmer has any sense. Altman can't seem to entertain the idea that many jobs are largely made up and unnecessary and people don't want to do them. Rather than looking to reduce or even eliminate work, he imagines endless fabrications to keep everyone busy for another 1000 years.
Graeber's notion of bullshit jobs was his answer to the question of why, after almost a century of technological progress, John Maynard Keynes's utopian vision for a 15-hour work week never eventuated.55 Keynes, J.M. (1930). "Economic Possibilities for our Grandchildren." In Essays in Persuasion (1931). Macmillan. Keynes predicted that by 2030, technological gains would reduce the working week to around 15 hours. For Graeber, instead of slowly removing human labour from the economy and freeing up time for everyone, we invented new ways to keep people busy. Graeber was, somewhat mischievously, calling bullshit on our present and reminding us of a future that never arrived.
Our AI future is already here. Even if we wilfully ignore the staggering environmental damage, the military use, the TechBro fantasies, and keep strictly to how AI is being used for mundane office work — let's be honest, a lot of it is for the outsourcing of bullshit tasks. In an ideal world, we could all have an honest conversation about this and agree to withdraw from the bullshit economy entirely. Imagine the resources we could save? Until then, the next time you come across AI generated content at work, rest assured, it's likely the person responsible didn't want it to exist either.
Notes