HALTUNG: On Developing Cultural Intelligence in the Age of AI

On style, AI and decision making.

"Natural taste is not a theoretical knowledge; it's a quick and exquisite application of rules which we do not even know."
— Montesquieu, 1759

No, I did not start reading up on French philosophers these days. But I did start using Readwise more again. Now that they have this ability to chat with the highlights one has created, there is suddenly a purpose to all these incisive digital highlighting. So far, it was hard to make proper use of them. Now, using the powers provided by a RAG systems (more on that in a minute), Readwise manages to allow me to query what I was highlighting even if I don't know anymore what this passage exactly caught my eye.

So where is this quote from? It's – well of course – from one of Kyle Chayka's essays. This one was published at a now defunct VOX Media publication called Rackd in 2018 and had the flavor of being ahead of the conversation by asking weather algorithms having destroyed our taste. I mean your taste. Definitely not my taste. You can access the full article through my readwise "hard" carbon copy.

But back to French philosophers: What if we understood personal style not as mere preference or aesthetic choice, but as what Germans call "HALTUNG" (can you even say a German word without it sounding like someone is screaming it at you?) — a term that encompasses stance, bearing, and attitude, but suggests something deeper than any of these English approximations can capture? “HALTUNG” implies a way of being that allows for instantaneous, culturally informed judgments. It's not just what you choose, but how you choose. Not just what you know, but how you know it.

"HALTUNG" implies a way of being that allows for instantaneous, culturally informed judgments. It's not just what you choose, but how you choose."

(Sidebar: Notice how I manage to talk about Germans in a way that acknowledges my German citizenship while simultaneously positioning myself as an observer—a perfect illustration of the third culture perspective that allows me to both belong and analyze from a distance. This dual positioning is precisely the vantage point that informs my understanding of “HALTUNG” as something both inherited and consciously cultivated.)

On Cultural Intelligence as System

In technical terms, we might think of “HALTUNG” as a sophisticated form of cultural intelligence — something akin to having your own, highly personalized Retrieval-Augmented Generation (RAG) system. But what exactly is a RAG system, you ask? Unlike conventional LLMs (Large Language Models) that generate outputs solely from their training data, RAG systems first retrieve relevant information from a separate, curated knowledge base before generating responses.

Consider what happens when someone asks ChatGPT or Claude about Jim Jarmusch's film "Ghost Dog: The Way of the Samurai." With a standard LLM approach, the model simply generates a response based on patterns it learned during training. But with a RAG system, the AI first retrieves specific information about the film's integration of the samurai code, hip-hop aesthetics, and mafia narratives, then uses that retrieved information to craft a more precise response. Forest Whitaker's character religiously follows an ancient samurai code while operating within modern urban culture, demonstrating how rule-following and rule-breaking coexist in cultural intelligence. The difference in AI terms is like comparing someone speaking entirely from general knowledge versus someone who quickly references specific sources before responding.

Artist Sougwen Chung's robotic collaboration system D.O.U.G. demonstrates this distinction beautifully: rather than generating art from scratch like Midjourney, it draws from a carefully curated archive of the artist's 20-year painting history to maintain stylistic continuity and artistic coherence.

"Our human 'cultural RAG' draws not just from data points but from lived experience, embodied knowledge, and cultural context."

This difference between retrieval-based and pure generation systems helps us understand how human taste operates. Like Chung's robots accessing her artistic archive, our cultural intelligence draws from a personal database of experiences, references, and patterns. But while AI RAG systems retrieve and process information through strict computational rules, our human "cultural RAG" draws not just from data points but from lived experience, embodied knowledge, and cultural context. Most importantly, we can dynamically adjust how we access and apply this knowledge based on subtle contextual cues that AI systems might miss entirely. Most importantly, it possesses a capability that current AI systems struggle to replicate: the ability to break its own rules when culturally appropriate.

On Breaking Rules by Knowing Them

This capacity for meaningful rule-breaking emerges from a deep understanding of cultural systems themselves. Fashion designer and cultural polymath Virgil Abloh exemplifies this understanding perfectly. As the first Black artistic director of Louis Vuitton's menswear collection and founder of Off-White, Abloh reshaped contemporary fashion by working at the intersection of streetwear, luxury, architecture, and fine art. His influence extended far beyond clothing, fundamentally altering how culture understands the boundaries between these previously siloed domains.

Central to Abloh’s approach was his ‘3% rule’ – a philosophy positing that strategic, minimal interventions (like repositioning logos or adding ironic text) could transform existing designs into culturally significant statements. While formally trained in architecture, Abloh subverted traditional design education by focusing on cultural resonance over originality. His Pyrex Vision project exemplified this: screen-printing ‘Pyrex 23’ (nodding to Michael Jordan and street culture) on deadstock Ralph Lauren flannels sold at luxury markups became a commentary on consumerism and authorship. The brand’s provocation – blending Renaissance art with streetwear – demonstrated how cultural understanding could amplify even minor design tweaks into viral phenomena.

Abloh's ability to identify which 3% to change stemmed from specific forms of cultural literacy. He possessed deep historical knowledge of fashion lineages, understood the semiotic value of cultural symbols across different communities, and recognized how context transforms meaning. When he placed "Pyrex 23" on Ralph Lauren flannel, he was deliberately invoking multiple referents simultaneously: basketball culture (Michael Jordan's jersey number), street drug culture (Pyrex glass used in cooking crack cocaine), and high-ish fashion (the quality of the base garment). This wasn't just intuition—it was a precise calculation based on understanding how these references would resonate across different cultural spheres.

On the Efficiency of Cultural Judgment

This is where the power of “HALTUNG” in an AI age becomes clear. While computational algorithms excel at pattern recognition and replication, they operate within strict logical constraints. Human taste, by contrast, functions as a self-aware system that understands not just patterns but context. Not just rules but their cultural significance.

This returns us to Montesquieu's insight about "rules which we do not even know." While we possess our own personal cultural RAG system, we don't always consciously understand how it operates. The countless experiences, cultural exposures, and contextual learnings that inform our taste occur at scales too vast and subtle for complete self-awareness. Yet what we can assert with confidence is that deliberately immersing ourselves in diverse cultures, contexts, and perspectives enriches this system. We accumulate knowledge that is simultaneously tangible and intangible, explicit and implicit, all of which becomes accessible through our personal cultural RAG when the moment demands it. The "unknowingness" Montesquieu identified isn't a limitation but rather the compressed efficiency of cultural intelligence operating at its highest level.

"The 'unknowingness' Montesquieu identified isn't a limitation but rather the compressed efficiency of cultural intelligence operating at its highest level."

The efficiency of this system is remarkable. Like a well-trained neural network, developed taste allows for nearly instantaneous decision-making. But unlike AI systems, which must calculate every possibility, human taste operates through what Montesquieu called "quick and exquisite application of rules which we do not even know." This unknowingness isn't a bug; it's a feature. It represents a form of compressed cultural intelligence that combines the speed of algorithmic processing with the depth of human understanding.

(I highly recommend reading everything by Vaughn Tan on not-knowing.)

On Developing “HALTUNG” in an Algorithmic Age

As our world becomes increasingly mediated by algorithmic systems, this capacity for cultural algorithmic thinking becomes more crucial, not less. Developing “HALTUNG” isn't just about standing out in an AI-driven world; it's about developing a sophisticated system for processing culture that operates on principles AI can't replicate (but for some it can mimic it). It's about building a personal cultural intelligence that can move seamlessly between pattern recognition and pattern breaking, between rule-following and rule-bending.

The future of personal style lies not in competing with AI's computational capabilities but in developing our uniquely human capacity for cultural intelligence. Our "cultural RAG" systems — our ability to process and respond to culture through developed taste — represent not an outdated approach to be replaced, but a sophisticated form of intelligence that becomes more valuable precisely because it operates on principles that can't be reduced to pure computation.

"It's not about having better taste than AI; it's about having a different kind of taste altogether."

It's not about having better taste than AI; it's about having a different kind of taste altogether — one that combines the efficiency of algorithmic processing with the depth and flexibility of cultural understanding. As someone who has navigated the spaces between cultures, I've come to appreciate that this flexibility—this capacity to see both the system and its exceptions—may be the most valuable intelligence we can cultivate.