Beyond “Critical Thinking”: Five Epistemic Functions for Learning With Artificial Intelligence

Artificial intelligence has intensified a question education has long tried to keep quiet: what do we mean by “thinking well,” and who gets to decide?

Students now encounter knowledge through systems that generate explanations, summaries, images, and arguments with extraordinary fluency. These systems do not simply deliver information. They shape what appears plausible, legitimate, and worth attending to. In response, educators often reach for “critical thinking” as a safeguard—as if it were a neutral, universal skill that can be applied anywhere, to anything.

But critical thinking is not a single, universal inheritance. It is one historically situated label—shaped largely through Western schooling and philosophy—applied to a family of epistemic practices that exist across cultures and histories under different names, values, and accountabilities. Many of these practices long predate Western philosophy; many have persisted despite its dominance, often in forms rendered marginal or invisible by colonial education systems (Battiste, 2013; Smith, 2012).

Rather than defending or discarding “critical thinking,” this article reframes thinking well through five epistemic functions that recur across human traditions of knowing. These functions are not techniques to master, but ways of relating to knowledge—ways that artificial intelligence both amplifies and distorts.

A note on scope and care: the traditions named here are not presented as neat or unified categories. Within Indigenous, Buddhist, Confucian, and Western traditions alike, there is plurality, debate, and evolution. Examples are illustrative, not exhaustive. Where learning engages Indigenous knowledge, local community protocols—not generalized frameworks—must guide ethical practice (Archibald, 2008; Kovach, 2009; Wilson, 2008).

1. Discernment

Knowing What Kind of Knowledge This Is

Across many traditions, thinking well begins not with judgment but with discernment: recognizing what kind of knowledge one is encountering, what it can legitimately do, and what it cannot.

In Western analytic traditions, discernment often appears as evaluating claims, evidence, and reasoning. Frameworks such as Paul and Elder’s RED model—recognizing assumptions, evaluating arguments, and drawing conclusions—are one expression of this function. They remain useful, especially for slowing down AI outputs that sound authoritative while masking weak evidence or hidden assumptions (Paul & Elder, 2006, 2014).

But discernment is not unique to analytic reasoning. Many Indigenous scholars emphasize that knowledge is inseparable from relationship and obligation—and that what counts as knowledge shifts with context, purpose, and accountability (Wilson, 2008). Decolonizing education, in this sense, involves refusing Eurocentric claims of universality and attending to what dominant institutions have rendered illegible or irrelevant (Battiste, 2013; Smith, 2012).

Other traditions frame discernment as distinguishing conceptual fluency from insight. Buddhist epistemologies, for example, often emphasize the limits of verbal explanation and the need for practices that cultivate wisdom beyond language (Gethin, 1998).

Across these approaches, a shared question emerges:

What kind of knowing is this—and what kind of knowing is it not?

AI routinely collapses these distinctions. Procedural instruction, moral guidance, historical summary, and lived experience arrive in the same grammatical form, creating the illusion that all domains of knowing are interchangeable outputs (Bender et al., 2021).

2. Interrogation

Testing the Limits of a Way of Knowing

Critical traditions everywhere include practices of interrogation—not merely challenging claims, but testing where a way of knowing stops being trustworthy or begins doing harm.

In Western philosophy, interrogation is often associated with Socratic questioning: probing assumptions, contradictions, and implications through disciplined dialogue. This remains valuable when AI produces coherent answers that tempt students to confuse fluency with reliability (Bender et al., 2021).

Yet interrogation is not always adversarial. Jo-Ann Archibald’s work on Indigenous storywork shows how knowledge can be tested and deepened through story, retelling, and relational listening—where meaning emerges through responsibility and community guidance rather than conquest-style debate (Archibald, 2008). Other Indigenous scholars emphasize resurgence through relational integrity rather than extractive critique (Simpson, 2017).

Still other traditions interrogate by noticing where language fails. Daoist philosophy repeatedly insists that reality exceeds linguistic capture, cautioning against forcing the world into definitive statements (Ames & Hall, 2003).

Across these traditions, the shared question is:

Where does this way of knowing stop being reliable—or become harmful?

AI systems cannot ask this about themselves. They do not know when they are overreaching. Interrogation, therefore, must be taught as an epistemic responsibility—not a rhetorical style.

3. Accountability

Knowledge as Something You Are Answerable For

In many dominant models of critical thinking, ethics is treated as an extension—something to consider after analysis. Yet across many epistemic traditions, accountability is constitutive of knowledge itself.

Shawn Wilson’s formulation—“research is ceremony”—captures this succinctly: knowledge is accountable to relationships, not merely to method (Wilson, 2008). Kovach similarly emphasizes that Indigenous methodologies require relational integrity and responsibility, not extractive data practices (Kovach, 2009). Kimmerer’s writing bridges Indigenous teachings and scientific inquiry through reciprocity as epistemic discipline, not sentiment (Kimmerer, 2013).

Feminist and Black feminist epistemologies echo this insistence. Standpoint theories challenge the fantasy of neutral, context-free observation (Harding, 1991; Haraway, 1988), while Black feminist thought foregrounds how power shapes whose experiences count as evidence and whose knowledge is treated as authoritative (Collins, 2000).

AI exposes this gap starkly. Algorithmic systems generate knowledge without consequence, authorship, or obligation. They are not accountable for harm, yet they routinely reproduce inequity by reflecting biased data and discriminatory infrastructures (Noble, 2018).

The critical question becomes:

Who bears the consequences if this knowledge is believed, shared, or acted upon?

Learning to refuse knowledge that arrives without responsibility may be one of the most urgent epistemic capacities in an AI era.

4. Situatedness

Knowing From Somewhere, Not Everywhere

No serious tradition treats knowledge as placeless. Yet Western schooling has often aspired to a “view from nowhere”—a posture of neutrality that conceals its own location and interests.

Within Western traditions themselves, scholars such as Haraway argue for situated knowledges, insisting that all seeing is partial and accountable (Haraway, 1988). Harding similarly reframes objectivity as a strengthened practice grounded in recognizing standpoint and power (Harding, 1991).

Indigenous scholarship has long foregrounded situatedness through place, relationship, and protocol—claims historically dismissed by colonial institutions as “subjective” or “local” (Archibald, 2008; Battiste, 2013; Wilson, 2008). Other traditions, such as Confucian role ethics, emphasize moral reasoning as relational and cultivated through responsibility rather than abstraction (Rosemont & Ames, 2016).

AI appears to speak from everywhere at once. This produces an illusion of universality that masks its actual origins in particular training histories, institutional priorities, and economic incentives (Floridi, 2019; Bender et al., 2021).

The guiding question:

From where is this knowledge speaking—and where can it not speak from?

5. Restraint

Knowing When Not to Know, Say, or Conclude

Perhaps the most neglected epistemic function in modern schooling is restraint. Yet across wisdom traditions, knowing when to withhold judgment, delay conclusion, or refuse an answer is treated as discipline, not weakness.

Many Indigenous knowledge systems regulate when and how knowledge may be shared; some knowledge is conditional, relational, or not universally available (Kovach, 2009; Wilson, 2008). Daoist philosophy warns that excessive certainty can enact epistemic violence by forcing the world into categories that stabilize the knower (Ames & Hall, 2003). Buddhist traditions similarly caution against attachment to conceptual certainty (Gethin, 1998).

AI cannot practice restraint. It produces answers endlessly and instantly. If education rewards speed and certainty, AI will quietly become the model student.

Teaching restraint means cultivating the capacity to say:

I don’t know yet. This question may be poorly framed. This answer may be unsafe.

That is epistemic discipline.

Learning With AI as Epistemic Practice

Reframed through these five functions, learning with AI becomes less about using tools efficiently and more about living responsibly with knowledge.

This is not technical AI literacy. It is epistemic education.

Toward Epistemic Humility

What Western schooling often calls “critical thinking” is one name for a much older and wider human project: learning to live wisely with knowledge.

Artificial intelligence does not threaten that project by thinking too well. It threatens it by producing fluent knowledge without discernment, accountability, situatedness, or restraint—and by tempting humans to do the same.

The task of education is not to defend a canon, nor to romanticize alternatives, but to cultivate epistemic humility: the capacity to recognize the limits of our systems, our traditions, and ourselves—and to remain responsible within those limits (Floridi, 2019).

That work did not begin in the West. And it cannot survive there alone.

References

Ames, R. T., & Hall, D. L. (2003). Dao de jing: Making this life significant. Ballantine Books.

Archibald, J. (2008). Indigenous storywork: Educating the heart, mind, body, and spirit. UBC Press.

Battiste, M. (2013). Decolonizing education: Nourishing the learning spirit. Purich Publishing.

Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). On the dangers of stochastic parrots: Can language models be too big? 🦜 Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, 610–623. https://doi.org/10.1145/3442188.3445922

Collins, P. H. (2000). Black feminist thought: Knowledge, consciousness, and the politics of empowerment (2nd ed.). Routledge.

Floridi, L. (2019). The logic of information: A theory of philosophy as conceptual design. Oxford University Press.

Gethin, R. (1998). The foundations of Buddhism. Oxford University Press.

Harding, S. (1991). Whose science? Whose knowledge? Thinking from women’s lives. Cornell University Press.

Haraway, D. (1988). Situated knowledges: The science question in feminism and the privilege of partial perspective. Feminist Studies, 14(3), 575–599. https://doi.org/10.2307/3178066

Kimmerer, R. W. (2013). Braiding sweetgrass: Indigenous wisdom, scientific knowledge, and the teachings of plants. Milkweed Editions.

Kovach, M. (2009). Indigenous methodologies: Characteristics, conversations, and contexts. University of Toronto Press.

Noble, S. U. (2018). Algorithms of oppression: How search engines reinforce racism. NYU Press.

Paul, R., & Elder, L. (2006). The miniature guide to critical thinking: Concepts and tools. Foundation for Critical Thinking.

Paul, R., & Elder, L. (2014). Critical thinking: Tools for taking charge of your professional and personal life (2nd ed.). Pearson.

Rosemont, H., Jr., & Ames, R. T. (2016). Confucian role ethics: A moral vision for the 21st century? Oxford University Press.

Simpson, L. B. (2017). As we have always done: Indigenous freedom through radical resistance. University of Minnesota Press.

Smith, L. T. (2012). Decolonizing methodologies: Research and Indigenous peoples (2nd ed.). Zed Books.

Wilson, S. (2008). Research is ceremony: Indigenous research methods. Fernwood Publishing.