Should You Be Weird?

In a recent interview, Palantir CEO Alex Karp said that he believes that AI will “destroy humanities jobs.” A philosophy graduate himself, Karp is pessimistic about the long-term employability of humanities grads, given that they tend to be “generalists” who specialize in critical thinking and problem solving, things he thinks AI is better suited for. Those who do have a future, according to Karp, are those who either have “vocational training” or are, in his words, “neurodivergent.”

Ironically, Palantir is arguably in need of people who specialize in the humanities, especially ethics. Regardless, Karp’s advice seems to be going against the grain: numerous other CEOs and economists have stated recently that the so-called “age of AI” has led them to look to liberal arts majors as sources of creativity that is outside AI’s wheelhouse. As one researcher puts it, there is currently a “weirdness premium” when it comes to job applicants.

While Karp and his critics disagree about the future job-readiness of humanities graduates, both seem to agree that there is something important about being weird. But what kind of weirdness is worth pursuing? Should you, too, be weird?

Weirdness might be considered particularly valuable at this moment in time because what AI produces – specifically, generative AI programs like bog-standard chatbots – is decidedly not weird. Or, if it is weird, it is weird by accident: AI-generated images and videos can often “look weird” by being uncanny or unnatural, and the output of chatbots can seem “weird” when they produce nonsense. But this weirdness is the result of AI programs not working properly: when they work as they should, they produce content that tends to be perceived as generic. This is a consequence of how these programs work: chatbots produce content by finding the most frequent combinations of parts of words in their database and strings them together. As a result, chatbots tend to produce content that is the average of the most common ideas, and is thus decidedly not weird.

If AI isn’t giving us the weirdness we want, where should we find it? “Weird” can be used in lots of different ways, depending on the context: we might think that a person is weird because of some facet of their personality we don’t really like or get; unexpected outcomes of actions or experiments are sometimes described as weird; something can be weird to our senses (tasting weird, smelling weird, etc.); art can be weird in plethora ways; etc. And, of course, weirdness is dependent on context: different people from different places will consider different things weird.

While many of these senses of “weird” are pejorative, the kind of premium on weirdness is presumably not that kind of weird we talk about when we refer to people that “weird us out” or when our yogurt tastes weird. What we’re looking for, then, is a kind of virtuous weirdness.

Although we use the word in lots of different ways, we can identify some characteristic aspects of what it means to be weird. Weirdness seems to involve something unusual, in the sense of being both infrequent and contrary to our expectations; after all, if everyone was doing it, and it wasn’t surprising, it wouldn’t be weird. Weirdness also seems to require an aspect of deviance. A deviant act is one that is recognizable as a type, but goes against the grain in some way. For example, if I were to say that my yogurt tastes weird, it is something that I would still recognize as being yogurt, just different, in some way. As such, weirdness is not chaos, randomness, or mere noise: it does not completely abandon the structures that we are accustomed to, but rather pushes on them in a way that causes us to reconsider how we interact with those structures.

The conditions of being unusual and deviant need to be present at the same time for something to be weird. Being unusual by itself is not necessarily weird: for example, we might meet someone who has the relatively niche hobby of stamp collecting, and this might be surprising given what we know about their personality, but this isn’t enough to say that they are weird. Similarly, mere deviance does not make for weirdness: there are plenty of behaviors that are “deviant” in the face of certain common norms, e.g. petty crimes like shoplifting, graffitiing, or doing drugs in places where that’s not allowed. We might argue about the morality of such acts, but they certainly aren’t “weird.”

Of course, acting in ways that meet all the conditions for weirdness does not necessarily result in anything we would want to aspire to. A virtuous weirdness might then require being directed towards a worthwhile goal. Consider, for example, a contemporary paradigm case of virtuous weirdness: the Quebecois band Angine de Poitrine. They have taken the internet by storm precisely because they are so weird: the two-person (or potentially alien) band, dressed in black-and-white spotted outfits and sporting proboscises of varying rigidity, have earned them the title of “the world’s weirdest party band.” The band has sometimes been called “anti-AI”: whereas AI-created music often sounds like an anodyne slurry of only the safest radio hits, Angine de Poitrine is surprising and captivating, even if the music itself is not really your thing.

This kind of weirdness is unusual – no one else is really doing what Angine de Poitrine are doing – and it is deviant: the music they create is different, but not so divergent from other music as to sound like mere noise. It is also directed at what seems like a worthwhile end: as an artistic expression, and to make us think about music in a way that we may not have considered before.

This is the kind of weirdness that cannot be replicated by AI, and is at a premium among today’s CEOs – although they are likely looking for weirdness that is directed towards a different end, namely, that of generating solutions to complex problems. Whether this kind of weirdness is something that can be cultivated by an education in the humanities or whether you simply need to be born with it is up for debate.

So: should you be weird? There are practical reasons why you might want to be weird, since you might be able to do things that AI can’t or won’t do, so your future employment opportunities will be safer. But there seems to be something antithetical about weirdness that is cultivated for the sake of impressing tech CEOs who are bored with their AI. Indeed, AI companies are frequently in search of those with niche expertise in creative disciplines to train their AI models, presumably with the goal of making them better at producing an ersatz weirdness that distinguishes them from their competitors, at least in some small way. But weirdness that is used to supplement or train AI then risks being self-defeating, making the unusual and surprising usual and predictable. It doesn’t seem like virtuous weirdness can be directed towards its own destruction.

That’s not to say that the only virtuous weirdos are those who, say, make weird art that they will never profit from, and weirdness may indeed be on display when coming up with creative new ways to solve problems in areas that many people would find mundane. But virtuous weirdness, no matter where it’s exhibited, seems to require a love of the game, valuing its unusualness and deviance for their own sake, at least in part. While we should rightly value weirdness, if your sole motivation for being weird is that you can secure a job at Palantir, then you’re probably not being the good kind of weird.