When You Can’t Know for Sure:

Charles Sanders Peirce and the Problem of Epistemic Anxiety in the Age of AI

Something has shifted in how it feels to hold a belief. Not in a single moment of rupture, but gradually, the texture of knowing has changed. You encounter a claim, find it plausible, repeat it, and then something nags. Was that real? Did I just pass along something that was generated, confabulated, assembled from statistical patterns with no referent in the world? The question doesn’t resolve cleanly. You move on, but the ground feels slightly less solid than it did before.

Exploring this through phenomenologically can be helpful because it’s genuinely new. It’s not the old fear of being wrong, that’s familiar enough and part of intellectual life. It’s something stranger: a vertigo that comes from the breakdown of the informal cues we’ve always used to navigate credibility. Tone, apparent expertise, the feel of a source – these have always been imperfect proxies, but they were ours, developed over decades of reading and listening and triangulating. AI-generated content has made them unreliable in a new way, because it can reproduce all the surface features of credibility without any of the epistemic grounding that was supposed to justify them.

What emerges from this isn’t just skepticism. It’s a kind of chronic unease around the act of believing itself, a difficulty committing to positions, a hypervigilance around sources, sometimes a paralysis around communication. The verification ideal that underpins modern intellectual life starts to feel both more urgent and more impossible simultaneously. That combination is its own particular form of suffering.

Peirce, I want to argue, has something useful to say here. Not as a solution (philosophy rarely offers those) but as a reorientation that changes what question you’re actually trying to answer.

The Verification Ideal and Why It’s Failing Us

There’s a tacit standard operating underneath epistemic anxiety – call it the verification ideal – which holds that responsible belief requires confident access to ground truth. If I can’t verify the source, I shouldn’t believe it. If I believe something that turns out to be false, I’ve failed epistemically, maybe even morally.

This ideal has a long philosophical history, and in many contexts it’s appropriate. As an operating standard for daily cognition in 2026 however, it’s generating a particular kind of suffering, because it sets up a demand the environment can no longer meet. The amount of information a person encounters in a day is too large to verify at source. With AI-generated content now mixed into the information environment without reliable markers, that verification task has become in many cases technically impossible. The signals we’d use to assess it have been hollowed out.

When you hold the verification ideal as a psychological standard and simultaneously live in an environment where verification is frequently unavailable, you’re going to experience chronic unease. That’s not a character flaw or a cognitive failure; it’s what happens when an internal standard and external reality are genuinely mismatched. The anxiety is, in a strange way, the appropriate response to an impossible demand. Which suggests the demand itself might need examining.

Peirce’s Account of Belief

This is where Peirce becomes interesting. In his 1877 essay “The Fixation of Belief,” Peirce proposed that beliefs aren’t mirror-images of reality, they’re habits of action. To believe something is to be disposed to act in certain ways. Belief is functional, not representational. This is a significant shift in what’s actually at stake when you hold a belief. On the standard picture, a false belief is bad because it misrepresents reality, it’s an error in your mental model of the world. On Peirce’s picture, a false belief is bad because it disposes you to act in ways that won’t work. It’s a practical failure rather than a representational one, and that distinction matters more than it might initially seem.

The philosophical implication reorients the central question from “is this true?” to “what does holding this belief commit me to doing?”

That’s a genuinely different question, and in most cases a less vertiginous one. The verification question can be unanswerable; you can spiral into it indefinitely without traction. The pragmatic question is usually more tractable, because you can actually examine what behavioural and interpretive commitments a belief involves and whether those commitments seem reasonable given what you do know.

Imagine, for example, that you’ve read something, repeated it, and then discovered the source was unreliable. The verification-ideal version of this experience is: I believed and spread something false, which means I’ve failed at truth-seeking, which implicates something deep about my relationship to knowledge. It’s identity-level, hard to exit. The Peircean version is: what was I disposed to do based on this belief? Did acting on it cause harm, create a false impression, commit me to positions I’d need to revise? That’s a concrete, examinable question, one that sometimes reveals a real problem worth addressing and sometimes reveals the stakes were lower than the anxiety suggested. Either way, you’re working with something rather than chasing your own tail.

Fallibilism as Intellectual Honesty

Peirce also developed what he called fallibilism which is the position that we act on the best evidence available while acknowledging we might be wrong. He was clear that this wasn’t skepticism; it wasn’t a philosophical counsel of despair. It was simply an honest description of how inquiry actually works. We form beliefs, act on them, encounter results, revise when needed. Certainty isn’t the operating standard; calibrated, revisable confidence is. This maps onto how cognition actually functions rather better than the verification ideal does. In practice we don’t verify things from first principles before holding them. We operate on trust, on source reliability heuristics, on contextual coherence, on background assumptions we’ve never explicitly examined. That’s not intellectual laziness it’s just how cognition has to work at scale. Peirce’s fallibilism unapologetically acknowledges that.

The verification ideal, by contrast, sets up a standard no one actually meets and then generates something like guilt when people notice the gap. There’s something almost generous about fallibilism as a philosophical reframe as it offers: you were doing what all thinking beings do, acting on available evidence with finite resources. The question isn’t whether you could have been more certain; it’s whether you remain genuinely responsive to revision when better evidence arrives. That’s a more workable criterion, and arguably a more honest account of epistemic responsibility than the verification ideal ever was.

The Implicit Pragmatism Already Embedded in How We Cope

What makes this more than a philosophical detour is that pragmatist thinking is already operative in how reflective people actually navigate information uncertainty, it just hasn’t been made explicit. When someone works through a troubling belief carefully, they’re rarely asking “does this perfectly represent reality?” in a strict philosophical sense. They’re asking something more like: is this useful to hold? Does it help me act in ways consistent with my values? What would change if I held a different one? What am I actually committed to by believing this?

Those are pragmatist questions. The move of examining evidence for and against a belief isn’t demanding certainty; it’s asking for a reasonable assessment in service of action. The downstream test is whether the belief is functional, livable, responsive to the world. This suggests Peirce’s framework could be used not just metaphorically but structurally when thinking through epistemic anxiety, as an explicit reorientation for moments when someone has become stuck on the unanswerable verification question. The shift would sound something like: we’re not going to fully resolve whether that source was reliable. Instead we can ask what you were disposed to do based on that belief, whether those dispositions were reasonable, and what you’d want to revise now, not because you got the truth catastrophically wrong, but because you’ve updated, which is exactly what intellectual honesty looks like.

What This Doesn’t Dissolve

Peirce doesn’t fix the structural conditions generating epistemic anxiety, and it would be dishonest to suggest otherwise. The information environment is genuinely degraded in ways that matter philosophically and practically, and a shift in how we hold beliefs doesn’t address the conditions producing that degradation.

There’s also a version of pragmatism that tips into something philosophically irresponsible, a pure instrumentalism about belief that says hold whatever works for you, which isn’t what Peirce meant and isn’t a position worth defending. Peirce cared deeply about truth; he just had a more sophisticated account of what the pursuit of truth actually looks like in practice. Fallibilism isn’t relativism. We might be wrong doesn’t mean anything goes. The reframe needs to hold both: loosening the grip of the verification ideal while preserving genuine care about accuracy and a real commitment to updating when the evidence warrants it.

Now for the experience of epistemic vertigo itself, the particular suffering of being unable to achieve certainty in an environment that structurally cannot provide it, Peirce’s pragmatism offers something livable. Not resolution, but a different relationship to the problem. The question isn’t whether you can know for sure. Often you can’t, and the environment we now inhabit has made that more nakedly obvious than it used to be. The question is what you’re prepared to do given what you have, and whether you’re genuinely open to revision.

That’s not a retreat from epistemic responsibility. It might be the most honest account of what epistemic responsibility has always actually required.