Metascience | Reality Bending Lab

Barely Knowing Anything: On Taste, Judgment, and Cognitive Elegance in the Age of AI

Fri, 29 May 2026 00:00:00 +0000

The clip above (Youtube Link) has become an internet meme used to mock out-of-touch executives (or PhD supervisors) who make consequential decisions without understanding or grasping the actual work involved. Rick Rubin, one of the most celebrated music producers in history, admits that he can barely play an instrument, cannot operate a soundboard, and knows nothing about music technically. The interviewer, visibly disbelieving, keeps pushing. What exactly is he being paid for?

“The confidence that I have in my taste,” Rubin answers, “and my ability to express what I feel.”

This answer sounds like the most sophisticated alibi for incompetence ever constructed. While I find it hilarious (the clip lives rent-free in my head), I’ve been recently realizing that it is also the most profound and prescient statement about the future of human cognition in the age of artificial intelligence.

Ingenium Is Dead. Long Live Elegantia.

We are living through a massive cognitive paradigm shift: the transition from an economy of generation to an economy of selection and curation. For the entirety of human history, the primary cognitive bottleneck was production. Writing an essay, coding a script, analyzing a dataset, painting an image: they all require substantial technical skill, sustained effort, and dedication. But artificial intelligence is driving the cost of solution (and even problem) generation to almost zero.

This is pushing the bottleneck for us humans elsewhere entirely. It is knowing whether that solution is actually any good. And harder still: knowing what the right problem was in the first place.

Cognitive Elegance

The skill that stands to become critical in this new landscape is something I want to call Cognitive Elegance. While the modern usage of elegance refers to something graceful or stylish, its Latin root, elegantia, actually derives from eligere (ex-: out) + legere (to gather, to read, to choose). Elegantia literally meant the faculty of choosing well: a blend of taste, clear vision, calibration, aesthetic sensitivity, and decisive judgment. The Romans paired it with ingenium: the creative genius. My argument is that the AI era is inverting their relative value. Ingenium is becoming cheap. Cognitive Elegance is becoming everything.

Cognitive Elegance is not expertise in any narrow domain. It is something more like a meta-faculty: the internalized capacity to sense quality, to distinguish signal from noise, to know with confidence when something is good; and to trust that judgment enough to act on it decisively.

A Problem for Education

From my work at the university, it is becoming painfully clear that Cognitive Elegance is precisely the skill most at risk in the current generation of students.

Faced with an infinite generative engine, the temptation is what Xu et al. (2025) have called cognitive agency surrender: the wholesale offloading of the intellectual process to an LLM, accepting its output without ever engaging an internal compass of quality. A striking recent study from MIT’s Media Lab (Kosmyna et al., 2025) provided neuroscientific evidence for this phenomenon: students who consistently used AI for writing tasks showed measurably “weaker brain connectivity” than those who wrote unaided, and 83% could not quote from essays they had just “written,” suggesting not just dependence, but a genuine dissolution of cognitive ownership.

And the problem might actually run deeper than blindly accepting flawed outputs. It is ceding navigation. When a student hands a question to a language model, they often surrender not just the answer but the shape of the problem itself, letting the model define what counts as a relevant question, which variables matter, and which direction to pursue. The result is not just mediocre output, but a systematically misdirected inquiry process, often leading to dead ends (students wasting hours arguing with their AI assistant, going down some rabbit hole because the initial problem was not correctly identified).

This is where Cognitive Elegance becomes existential. Without it, you cannot even recognize when you’ve been led astray.

Existing Shoulders to Stand On

Cognitive Elegance is not without precedent in the psychological and philosophical literature, though no single existing construct captures it in full.

Educational psychologists have made the most direct approach with the concept of evaluative judgment, defined by Tai et al. (2018) as “the capability to make decisions about the quality of work of oneself and others,” encompassing knowledge of what “good” looks like in a domain and the confidence to apply that standard to novel problems. This is the closest relative, but it was theorized within academic assessment contexts, not in the face of machine-generated ambiguity at scale. Metacognitive research contributes epistemic calibration, the alignment between how confident you are in a judgment and how accurate that judgment actually is. Aristotle’s moral psychology gives us phronesis (practical wisdom): the navigational capacity to identify the right problem before attempting to solve it.

Personality psychology highlights openness to aesthetic experience as a dispositional substrate that might enable this sensitivity in the first place. We must also consider the Need for Cognitive Closure (NFC), the dispositional desire for a firm answer and an aversion to ambiguity. A healthy tolerance for ambiguity, paired with the ability to decisively reach closure, is the very engine of clear vision. Broadening the scope, this touches upon aesthetic and moral judgment. In aesthetics, we call this “taste”: a formalized, internalized sensitivity to harmony and value. In moral psychology, it is moral reasoning. Cognitive Elegance unites these domains, acting as a generalized sensitivity to what is “right.”

Jung’s concept of Intuition adds a further dimension: not a vague feeling but a distinct perceptual mode that apprehends patterns and wholes directly, without deliberate analysis. This is the mechanism by which Cognitive Elegance often operates, surfacing as a gut feeling or an impression before it can be articulated. I believe it echoes the Sanskrit concept of Viveka, the faculty of discernment between the real and the illusory, signal and noise, that was considered the foundational intellectual virtue in Vedantic thought.

What Cognitive Elegance names is the integration of all these dimensions into a single operative capacity: one that is domain-general, hopefully trainable, and, as we are now discovering, alarmingly easy to atrophy.

What Education Must Now Confront

The critical question of modern education is no longer how to teach students to produce. It is whether the capacity for vision and clear selection can be actively cultivated, and whether our institutions are organized to do so. What we need, urgently, is to develop the capacity to measure Cognitive Elegance, and from there, to design pedagogical conditions that cultivate it.

Rick Rubin, pressed to justify his place in a recording studio, answered without embarrassment: confidence in his taste, and the ability to express it. He barely knows anything, except the one thing that matters: what is good and what is not. What I am not convinced of, however, is whether this ability is possible without a strong foundation of expertise. My gut feeling is that you cannot train Cognitive Elegance ex nihilo, without building up a substantial reservoir of domain-specific knowledge and experience to draw upon. But that is an empirical question, and one that we need to answer urgently.

References

Kosmyna, N., et al. (2025). Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task. arXiv:2506.08872 [preprint].
Xu, et al. (2025). Cognitive Agency Surrender: Defending Epistemic Sovereignty via Scaffolded AI Friction. arXiv:2603.21735.
Tai, J., Ajjawi, R., Boud, D., Dawson, P., & Panadero, E. (2018). Developing evaluative judgement: enabling students to make decisions about the quality of work. Higher Education, 76, 467–481.

From physique to intellect... and back again?

Mon, 03 Nov 2025 00:00:00 +0000

Welcome to the “Pub Theories on Dystopian Futures” series, where we engage in wild speculations about the future of academia and society.

TLDR; AI will commodify and devalue intelligence, leading to natural beauty becoming the next highly prized characteristic in Humans.

Physical health as primary desideratum

From a psycho-evolutionary perspective, many human behaviours can be understood as rooted in mating strategies. At their core, these involve signalling one’s genetic fitness to potential partners. In humans, this often manifests as displays of youthful beauty in women (as a proxy for fertility) and status in men (as a proxy for resource acquisition and protection). These sexual dimorphisms arise primarily from the biological asymmetry in parental investment between the sexes.

On top of these individual selective processes, there are also societal or group-level selective pressures, including accidental ones. For instance, the black plague that killed one in three Europeans in the 14th century exerted selective pressure on immune genes, notably increasing the frequency of protective allele variants (Klunk et al., 2022).

Importantly, at a population-level, pre-modern societies preferentially selected for physical characteristics like strength and health over intellectual ones. For hunter-gatherer men, strength and endurance were the best predictors of family provisioning and protection. Likewise, for the common farmer of the Feudal era, it didn’t matter whether you were smart or not; what mattered was whether you could labour and toil efficiently and survive the harsh conditions.

The rise of intelligence

With the Industrial Revolution, and overall advances in medicine, nutrition, and sanitation, life expectancy increased, and the population grew (e.g., the population of Britain increased five times in 150 years). Societies became more literate, relying on bureaucracies, trade and technologies, and increasingly meritocratic. This resulted in a shift in selective pressures: intelligence became a prime desirable trait. Due to these multiple converging factors (notably education, nutrition, and health), average IQ scores increased over the 20th century by about 3 points per decade (the so-called Flynn effect).

Today, IQ is one of the strongest single predictors of positive outcomes, including academic achievement, job performance, socioeconomic mobility, health, emotional stability, happiness, and even longevity (Lo, 2017). As such, being (perceived as) smart confers advantages in modern societies surpassing those of physical fitness.

Interestingly, the recent decades of data seem to provide some evidence for a possible stagnation or even reversal of the Flynn trend¹, suggesting we may be approaching a plateau… or a regime shift.

The devaluation of intelligence

Just as the agricultural revolution transformed society by lowering the need for less efficient food production methods, and just as the industrial revolution changed society by lowering the need for manual labour, the AI revolution is changing society by lowering the need for Human intelligence. If AI commodifies smartness, the competitive edge of human intelligence becomes diluted, relaxing related selective pressures. Once intelligence is cheaply available, the benefits of being smarter shrink, triggering a decrease in its value as a desirable status-marker trait.

In the near future, intelligence might no longer the main characteristic prized by society, providing positive outcomes and status. This raises the question: what comes next? What feature will serve as ground for future Humans to compete on?

Natural beauty and the emergence of a kalokagathos class

In a world where technology equalises for advantages related to cognitive and physical abilities and health, the new frontier of human competition may be aesthetic and embodied: beauty.

The ancient Greeks encapsulated this ideal in the term kalokagathos - the unity of the good, the true, and the beautiful within a single person. For them, beauty was not a shallow and vain superficial quality; it symbolised harmony, virtue, and authenticity.

As artificial intelligence, virtual realities, and algorithmic filters dominate experience, society may increasingly valorise naturalness - what appears genuine, unmediated, and unfiltered. The value of naturalness, authenticity, and directness² might come as a counter-movement to the rise of artificiality, syntheticity, and virtuality. In this context, the pendulum could swing towards an aesthetic moralism: beauty as truth and goodness. In this world, natural beauty might become the prime desirable trait, with all its paradoxical implications in the form of a post-AI society obsessed with fake naturalness - surgical enhancements and digital manipulations designed to look unaltered, and carefully manufactured authenticity.

The Kalokagathoi, living embodiment of beauty and virtue, would become the new elite class, dominating not by their power to do, but through their mere quality to be, unsullied by technological augmentation. This value shift towards natural reality might last… until our technological capabilities allow us to influence the most fundamental aspects of our biology via gene editing, synthetic biology, … But what comes after that is a story for another pub theory.

¹ The reversal of the Flynn effect has been documented in several developed countries, such as Norway, Finland, and the UK, with IQ declines ranging from 0.38 to 4.3 points per decade since the 1990s in some studies. However, this is not universal—gains continue in other regions—and causes are debated, including factors like changes in education quality, immigration patterns, environmental toxins, or even the rise of digital distractions like smartphones (Bratsberg & Rogeberg, 2018; Dutton et al., 2016).

² Directness refers to the idea of unmediated experience, e.g., accessibility of the source of an experience. This translates to direct experiences of nature, social interactions without digital mediation, and raw sensory experiences. This concept also applies to products, opposing hand-crafted goods and locally sourced food (for which the “source” and “origin” are directly accessible and known) to mass-produced items for which the creation process has been obscured and mediated through complex supply chains.

The AI revolution in academia: I see a silver lining for young scientists!

Mon, 08 Sep 2025 00:00:00 +0000

For many students and early-career researchers, the rise of AI is scary. The very skills they are spending years learning - writing, coding, summarising - are exactly the things AI is getting frighteningly good at. What, then, is the future of research careers in the age of AI?

While I don’t have a full answer to that question, I do think there may be a silver lining for young scientists. For decades, becoming an established “big shot” professor was associated with focusing on the “big ideas”, revelling in intellectual thinking and leaving the scientific grunt work to junior researchers. Therein lied the prestige: writing opinion pieces, commentaries, critiques and reviews. The glamour was in thinking, not doing.¹

In a fascinating twist, the rise of AI may disrupt this landscape. If AI excels at one thing, it is precisely writing, reviewing and summarising evidence, interpreting findings, and even formulating new hypotheses or planning experiments. What AI still cannot do, however, is roll up its sleeves and gather real-world data - the true backbone of empirical science². It can’t (yet) run studies, recruit participants, set up experiments, organise data management, or wrestle with messy datasets.

The centre of gravity in science may thus shift towards those who can do: Hands-on scientists, who might once have been relegated to the shadows, could see their skills and contributions gain new recognition. We may see less pressure to write endless papers and grants, and more emphasis on how the science was actually done: how data was collected, preprocessed, managed, and made accessible. Perhaps the introduction, discussion, and “key takeaways” sections of papers will become somewhat less important, while methods, results, and limitations gain greater prominence, allowing for more nuance and granularity³.

Whether this shift will make science better or worse is unclear. And that “silver lining” might end up as a “glorification of the grind” and a devaluation of the intellectual aspects of research, devolving the job of “Researcher” into technician work. But it might also reshape the academic landscape in a way that proves beneficial for young researchers and those who enjoy the practical aspects of research.

¹ Sure, thinking is important, and big shot professors are also doing a lot… ^_sometimes.

² Of note is following the “replication crisis” in psychology, a lot of voices have called for “more theory” and more “theorically-grounded research” (with the goal of cutting some of the nonsense out there). While theories are critical to guide data collection and interpretation, it is still the hard evidence that ultimately is the foundation of scientific knowledge.

³ One of the pervasive issue is that Humans have limited “context window” (~ working memory). When reading a paper, it is already very hard to keep track of all the results and details in mind and integrate them into a coherent picture. Moreover, with the increasing role of social media, science had to be made more communicable, digestible, punchy, and “sexy”. This has led to a tendency to oversimplify and overgeneralize findings. AI, with its ability to process and summarize large amounts of information, could perhaps help make more accurate and data-grounded interpretations and summaries. (it might be wishful thinking, but who knows!)