Before the prompt, there was the card
In 1975, before anyone had heard of “prompt engineering,” Brian Eno and Peter Schmidt created something called Oblique Strategies.
It wasn’t software. It wasn’t a machine. It was a deck of cards filled with short, cryptic instructions meant to help artists get unstuck.
These cards did something very similar to what modern AI prompts do: They introduced a little bit of randomness and uncertainty to spark new ideas.
This essay argues that Oblique Strategies was more than a quirky creative aid — it was a prototype for prompt engineering, a visionary artifact that prefigured the mechanics of generative systems. By exploring how Eno’s cards function in relation to both human creativity and machine intelligence, we can draw a historical and conceptual bridge between analog disruption and algorithmic generation.
We’ll explore what these systems share:
A logic of indirection;
An embrace of imperfect input;
A shared belief that creativity emerges from reframing the frame itself.
The oblique prompt – creative instruction as interpretation engine
An Oblique Strategy is a short, open-ended suggestion. Some examples:
“Use an unacceptable color.”
“Only one element of each kind.”
“Give way to your worst impulse.”
“Remove ambiguities and convert to specifics.”
“Accept advice.”
These are not answers. They are questions disguised as commands. They don’t tell you what to do — they force you to decide what they mean. In this way, the card becomes a kind of semantic mirror — what you see in it is a reflection of what you’re avoiding, hiding or resisting in the creative process.
Each Oblique Strategy is a prompt without a destination. Its job is to start a deviation, not finish a sentence.
In one recording session with David Bowie during Low, Eno famously pulled an Oblique Strategy card reading, “Honor thy error as a hidden intention.” Bowie had been frustrated by a mistake in the synthesizer track. But the card reframed the glitch as a gift. They kept the error, built around it, and that accidental texture became the signature of the song’s atmosphere.
It’s a vivid reminder that the difference between noise and innovation is often the willingness to re-see what you already have.
The prompt as catalyst – how LMs work with uncertainty
In the context of a large language model (LLM), a prompt serves as the seed of output. Whether it’s “write a bedtime story about a dragon who’s afraid of fire” or “summarize Kant like a surfer,” the prompt does not command — it suggests a probable continuation.
Prompt engineering, then, is the skill of designing that ambiguity with intention. The most effective prompts are not always the most detailed, but the most evocative — just enough to steer, not so much as to limit.
This is exactly what Oblique Strategies do.
They don’t offer data. They shape context. They don’t request logic. They invite mutation. They operate like mental latent space navigators, directing thought toward unfamiliar coordinates in the creative map.
Consider how specific Oblique Strategies map to modern AI prompts:
“Use an unacceptable color.” In prompting terms, you might ask: "Write a product description using the most over-the-top language you can imagine.” Both instructions invert the expected norms to force novelty.
“Give way to your worst impulse.” A parallel prompt could be:“Describe a utopian city through the perspective of its most corrupt official.” The prompt functions not by prescribing but by provoking.
“Remove ambiguities and convert to specifics.” This can resemble:“Rewrite this abstract mission statement in plain sixth-grade English.” Both interventions reframe rather than define.
Eno as early prompt engineer
Let’s take a moment to reframe Eno’s role. He wasn’t just a musician or producer. He was a systems thinker — a designer of inputs. His work across ambient music, generative compositions and creative philosophy is defined by a single principle:
“Set the conditions. Let the system evolve.”
This is, word-for-word, the foundational thinking behind gen AI. Eno was engineering emergence long before we had transformers and diffusion models. He wasn’t solving problems. He was provoking possibility.
His method of creating ambient music — setting up loops of different lengths on multiple tape machines and letting them phase against each other — bears an eerie resemblance to prompt + inference loops in AI systems today.
Eno’s tape loop technique was deceptively simple: He would record several short loops of audio on magnetic tape, each loop set to a slightly different length. As the loops played back simultaneously, they would phase in and out of sync, creating evolving textures that never precisely repeated. For example, one tape might be 10.2 seconds long, another 9.8 seconds.
This system introduced two critical dynamics:
Controlled randomness — the loops drifted unpredictably, making each moment emergent.
Generative complexity — from a handful of inputs, the permutations multiplied exponentially.
This process is strikingly similar to how transformer models generate text: A prompt seeds an initial state, and each probabilistic continuation slightly alters the trajectory, producing output that feels organic and alive.
In both cases, the creator becomes a curator of possibilities rather than an author of certainties.
Prompting as ritual – the psychology of structured disruption
What makes a good prompt?
Not clarity, but constraint.
Not specificity, but structure.
Not truth, but tension.
Oblique Strategies introduced this logic long before LLMs. The deck is finite, but the combinatorial possibilities it opens are nearly infinite, precisely because the instruction is interpretable. It feels like it’s been written just for you, in this moment, even though it’s a blind draw.
This mirrors the strange intimacy of LLM prompts: Users often describe “conversations” with the model that feel deeply personal, even mystical. But the magic isn’t in the model — it’s in the prompt as mirror.
Both Eno’s cards and AI prompts function as ritualized inputs into stochastic systems. The goal is not to control the output, but to initiate a transformation.
From card decks to code prompts – a continuum of creative tools
If we see creativity not as a burst of genius but as a series of structural perturbations, then Eno’s work and AI’s hallucinations begin to merge. The connective tissue is the prompt itself — not as instruction, but as an aesthetic and philosophical intervention.
We might chart this lineage as:
Era | Tool | Function |
Pre-digital | Shamanic ritual/trance | Induce altered states for symbolic recombination |
Analog | Oblique strategies | Introduce randomness into structured creation |
Digital | Prompt engineering for LLMs | Shape probabilistic models to generate novel outputs |
Each stage represents a refinement of how humans externalize the muse — first as gods, then as cards, now as code.
Conclusion: Prompting as artform, art as prompt
Today, as AI moves from novelty to necessity, the art of prompting is no longer optional — it’s becoming the new creative literacy.
Brian Eno’s Oblique Strategies offer a powerful lesson: The most transformative work doesn’t come from perfect plans, but from well-crafted invitations to the unknown.
In a world where algorithms increasingly shape our creative output, the quality of the prompt determines the shape of the future.
A prompt is not just a question.
It is a choice.
A gamble.
A deliberate risk.
And if we do it well, it’s how we stay human in the age of the machine.
Roberto Santellana is the senior managing director for digital creative and technology at Verizon Creative Marketing Group.
Welcome to the VentureBeat community!
Our guest posting program is where technical experts share insights and provide neutral, non-vested deep dives on AI, data infrastructure, cybersecurity and other cutting-edge technologies shaping the future of enterprise.
Read more from our guest post program — and check out our guidelines if you’re interested in contributing an article of your own!