AI and the Humanities' Motivated Reasoning Problem

I’d like to thank Victor Kumar for introducing me to these ideas and motivating me to write. I’d also like to thank Dan Williams, whose substack ideas directly influenced the content of this essay. I recommend everyone interested in motivated reasoning and misinformation read his works and subscribe to his page: Conspicuous Cognition.

Undergraduate humanities majors (and humanities academics) are pretty anti-AI (1,2,3). This shouldn’t be surprising, considering that the core of the humanities is the study of human culture and ideas. AI is not human whatsoever. It’s a language-producing model which gives us information based on how we prompt it… it’s lifeless… really. A large part of the humanities’ disdain for AI is warranted, specifically in the ways in which it’s affecting education. Ask any professor from history to anthropology and they can tell you all about the ways in which their students are abusing AI to get around work. For that matter, ask any undergraduate who uses AI to do their work, if they’re being honest, they’d probably admit that it’s rotting their brain, cognitively eroding them.

However, a good part of the humanities’ AI skepticism is completely ludicrous, and I think it exposes the ways in which many liberal arts experts and students suffer from elitist aesthetic preference and politically motivated reasoning. This was made strikingly apparent to me this past semester, when I was sitting in my weekly undergraduate Social Epistemology Seminar. The day’s topic was AI ethics, and we discussed both the negative and positive potentials of the new technology. More specifically, we thought of what a world would look like where AI has near-perfect logical reasoning or consistently good decision making. These possibilities may never come to fruition, although we have good reason to believe AI will be greatly beneficial in many areas (cancer research, natural disaster response, disability services, etc). Though, if AI did get even near the level of perfect reasoning or decision-making, wouldn’t that be good for humanity?

My class wasn’t so convinced. This was bizarre to me. Wouldn’t decision quality across domains, like law, policy, science, and even everyday life be raised? It seems as though even small improvements compound because better reasoning would improve the institutions that generate future decisions. AI could even scale expertise to everyone, becoming a tutor at a level higher than even a university could provide. It would make us happier and live longer, since it would help avoid cognitive failures, like bad inference and brittle planning, which make up a huge share of human suffering.

My class didn’t believe AI could ever be so useful, and I’m not necessarily convinced either. But, it certainly has the potential to help humanity, and thus it’s worth it to nurture the technology so that we can use it to our benefit, right?

What shocked me is that many in my class came to the conclusion that we shouldn’t concern ourselves with AI at all. In fact, most thought AI would lead to our demise. It’s a cold, lifeless, and soulless technology that knows nothing about what it means to be human, and therefore shouldn’t be anywhere near vital human decision making. AI can’t understand us, and we are going to look back in decades and agonize over the mistake we made to let it run our world. Essentially, AI is not authentic, crafted, or “human-made” and even if it has great benefits, if it were to be implemented significantly in everyday life, its mechanistic nature would detract from the quality of life in other meaningful ways.

However, this view to me is brazenly classist. It simply ignores the costs of rejecting these tools, which are unevenly distributed. People with money can still buy scarce human attention, like doctors or tutors, while people without money often can’t. I’m not saying that we should tell low income families to go to ChatGPT instead of the doctor, but it is true that in the case of AI being a better than average-human medical decision maker, it’s a no-brainer to consult it for advice. Wanting “human-only” production of goods, and the prestige of the un-automated are preferences easiest to maintain from a position of security. After all, the affluent can still purchase the scarce human labour they need regardless of AI or not. This classist take, to me, is ironic considering that many of those advocating against AI consider themselves to be champions of equity. The irony runs even deeper when we remember that humanities majors, on average, have higher median parental income than any other major.

Before I start handing out our moral verdicts, let me say that although the class took place at my private university with a tuition well above 90K USD per year, it is more than likely that many in the class weren’t wealthy themselves. Regardless, one can hold a classist view without being in an affluent class.

I think that this view on AI and technology is rampant among younger people in the liberal arts – and it’s a major problem. We cannot be against technology that measurably improves quality of life, and we cannot shut down technology before it has the opportunity to do exactly that. If the left and humanities scholars reject AI, then risk rejecting a tool that advances the very egalitarian principles which define them. Imagine if we shut down the machine manufacturing revolution because we didn’t like the fact that our mugs were made by machines instead of people – are you kidding me???

The humanities needs to rid itself of this elitist taste. It only further isolates it from the rest of the real world, as it refuses to engage with the technology and developments of the modern day.

Let me illustrate the tension a little clearer here:

1. AI has plausible welfare benefits, like fewer medical errors, access to tutors in any field, cheaper services, disability accommodations, etc.

2. The cost of rejecting AI tools is unevenly distributed, because people with money can still buy the human version of what AI can replace.

3. Aesthetic norms aren’t neutral and signal taste or virtue.

4. So, opposition to AI (assuming it’s truly helpful in the ways outlined) can be best explained as status-guarding rather than true moral concern, i.e. protecting cultural hierarchy of authenticity and artisanship even though automation might expand access and reduce suffering.

There are other plausible reasons to oppose AI, like worries about bias, safety, misuse, labor displacement, surveillance, human relationships, and the like. These concerns are real (although I am still very skeptical about many of them). However, status/aesthetic based opposition is very prevalent in the liberal arts academy —the disgust for automation is real— and I think it’s allowing us to override welfare considerations despite the potential gains AI may give us.

Now let’s move on to the second reason I think this anti-AI, anti-tech view is so prevalent in the humanities – politically motivated reasoning. It’s no secret that the humanities skew left. I’m saying this as a woke left-wing undergraduate myself! Over the last half decade or so, in the arts we’ve been constantly reminded about the ways in which right-wingers relentlessly defend patently false beliefs and viewpoints, and how it’s the educated left’s role to steer the country back on course.

Moreover, polarized and tribal are two words often used to characterize today’s world. Many different philosophers, media scientists, sociologists, and data scientists have taken cracks at explaining how crazy our world (and more specifically the right) has become. Most have pinned the blame on misinformation. There seems to be rampant fake news swirling online, consumed and believed by millions (I literally cannot tell you how many times I’ve read “post-truth era” since 2020 man). You’ll be shocked to find out that misinformation isn’t really the problem, it’s motivated reasoning. I won’t hash out why here, but you can take a look at any number of Dan Williams’ posts on misinformation to get an idea of why this claim against misinformation is plausible. Motivated reasoning, more specifically the political strain can be defined as follows:

“Where positions on some policy-relevant fact have assumed widespread recognition as a badge of membership within identity-defining affinity groups, individuals can be expected to selectively credit all manner of information in patterns consistent with the irrespective groups’ positions. The beliefs generated by this form of reasoning excite behavior that expresses individuals’ group identities. Such behavior protects their connection to others with whom they share communalties.” — The Politically Motivated Reasoning Paradigm, Part 1: What Politically Motivated Reasoning Is and How to Measure It DAN M. KAHAN

Politically motivated reasoning has often been diagnosed on the right, and for good reason. But, it’s not a right-wing monopoly. It’s not just conservatives who tend to associate what is “morally right” or “true” with what confers with their most deeply held political beliefs. The left, and more specifically left-wing humanities academics, have their own identity protecting proclivities, and I think AI is increasingly one of the objects around which these proclivities organize.

The academic left has fashioned itself as the group doing critique, protecting truth, and resisting propaganda. But the problem with this self-conception is that it makes it easy to misidentify what’s actually driving the belief formation in our polarized context. Many frame misinformation as the main difference between our two distinctive American political factions, but the deeper driver is identity-protective cognition, which is shaping how individuals reason politically. By overemphasizing false information and positioning itself as the arbiter of truth, the left has become prey to the very tribal reasoning it claims made conservatives crazy.

A symptom of motivated cognition is that people selectively trust sources that flatter whichever in-group they seek to please and doubt sources which threaten it. Once you notice this mechanism, you can see why AI is a perfect object for the left to grab onto. In many humanities contexts, and again often for good reason, AI is symbolically loaded. It’s going to take our jobs, it’s a corporate tool, it’s labour displacing, it’s an amplifier of policing, and it flattens culture. These points are extremely debatable and not settled, yet the left has largely considered them put to bed and defaulted to doomerism. In other words, AI has been deemed morally contaminated, a thing we as liberal arts enjoyers are supposed to oppose for cognitive incentives. The questions surrounding AI are therefore not what is true about AI and under what conditions can it be beneficial? but instead, in what ways can I signal that I am one of the good understanders of the true ways in which AI is existentially destructive? I suspect that this line of thinking is more akin to moral identity performance than rigorous moral analysis.

Finally, this explains many of the particular objections to AI despite its potential. Obviously, one wouldn’t outright say they are making an argument to fit into a group narrative, instead they produce cleaner rationalizations, for example that AI is soulless. Notice how a claim like this shifts away from any sort of debate about empirics like error rates or guardrails and shifts to spiritual hygiene. Once you debate authenticity, any empirical data can be dismissed as missing the forest for the trees. Even if AI saves lives, it still violates the maxim that humans must be central. This is where the politically motivated reasoning lies, because it insulates the identity stance from counter evidence.

Let me put my argument plainly:

1. In polarized environments, beliefs function as identity signals.

2. When beliefs function as identity signals, people selectively interpret evidence to protect their group membership.

3. In left-leaning humanities, anti-tech/corporate is a central identity.

4. AI has been absorbed into that marker, so that opposition to AI signals moral seriousness and solidarity.

5. Therefore, many anti-AI attitudes in the left, and left humanities are not the product of rigorous cost-benefit analysis but instead of politically motivated reasoning.

To conclude, let me stress a few things. AI is not harmless and skepticism isn’t irrational. We should absolutely have a healthy amount of anxiety towards all technology and I think the role the humanities plays in that is important. I also am not trying to dog on the humanities or undergraduates. We are all getting attacked by a bigger fish, the federal government, and the biggest loser in that is my generation, Gen Z, who will likely all be raised without knowing anything about meaning in life or ethics beyond government religious slop.

Now that my apologies are out of the way, my point was that I’ve noticed that many in my philosophy circle reject AI not because they’ve genuinely compared risks to benefits under realistic AI predictions, but because an endorsement of AI’s benefits constitutes a threat to their moral self-conception. It’s easier to condemn the tool than to do the hard work of specifying its constraints, demanding accountability, and figuring out where exactly its applications are that are welfare-improving. If the world is going to start taking what we do as philosophers and humanities scholars more seriously, then we need to look at ourselves in the mirror and deconstruct our most closely held convictions, in order to present as unbiased and not captured by elitist taste and tribal reasoning.