From cards to AI prompts
In 1975, musician and producer Brian Eno teamed up with artist Peter Schmidt to create a deck of cards called Oblique Strategies. These weren’t ordinary playing cards — each one carried a short, strange, often cryptic phrase designed to unlock creative blocks.
Some examples included:
- “Use an unacceptable color.”
- “Only one element of each kind.”
- “Give way to your worst impulse.”
- “Remove ambiguities and convert to specifics.”
- “Honor thy error as a hidden intention.”
These cards didn’t provide answers — they raised questions. They worked as a mirror, forcing the artist to see what they might be avoiding or suppressing.
In today’s digital world, AI prompts work in a strikingly similar way.
From Bowie’s studio to modern prompts
Oblique Strategies often turned accidents into creativity. For instance, during the recording of David Bowie’s album Low, a synthesizer error left him frustrated. Eno drew a card: “Honor thy error as a hidden intention.” The “mistake” became a defining element of the song’s atmosphere.
AI prompts function in much the same way — they don’t dictate, they nudge. For example:
- “Use an unacceptable color.” → “Write a product description using the most exaggerated language possible.”
- “Give way to your worst impulse.” → “Describe a utopian city through the eyes of its most corrupt official.”
- “Remove ambiguities and convert to specifics.” → “Rewrite this mission statement in plain sixth-grade English.”
In all these cases, the goal isn’t control, but provocation.
Tape loops and transformers: don’t control the system, set the stage
Eno’s philosophy was simple: “Set the conditions, let the system evolve.”
He experimented with magnetic tape loops of different lengths — one 10.2 seconds, another 9.8 seconds. As they played together, they slowly drifted out of sync, producing soundscapes that never repeated the same way twice.
This process revealed two dynamics:
- Controlled randomness — the loops moved unpredictably.
- Generative complexity — infinite textures emerged from just a few simple inputs.
Transformer-based AI models work in a remarkably similar way. A single prompt sets the initial state, and each probabilistic continuation shifts the trajectory, creating output that feels organic and alive.
Prompt as an artform
Eno wasn’t just a musician — he was a systems designer. His job was to shape the input conditions, then step back and let the system surprise him. That’s the very essence of generative AI: a prompt guides the model, and the system generates variations beyond the author’s full control.
As Eno himself put it:
- “Not clarity, but constraint.”
- “Not specificity, but structure.”
- “Not truth, but tension.”
A good prompt doesn’t deliver certainty — it triggers transformation.
A lineage of creative tools
Humanity has always looked for ways to guide creativity through structured disruption.
| Era | Tool | Function |
|---|---|---|
| Pre-digital | Rituals, trance | Alter states to recombine symbols |
| Analog | Oblique Strategies | Add randomness into structured creation |
| Digital | Prompt engineering | Steer probabilistic models to generate novel outputs |
This is a single continuum. What cards once did for artists, prompts now do for entrepreneurs, developers, and creators.
Conclusion: Prompting as creative literacy
AI is no longer a novelty — it’s becoming a necessity. In this shift, prompting is not just a technical trick; it’s a new form of creative literacy. Eno’s lesson still resonates: breakthroughs rarely come from perfect plans. They emerge from well-crafted invitations to uncertainty.
A prompt is not just a command.
It is:
- a choice,
- a risk,
- an opening to possibility.
If used wisely, prompting won’t just make AI useful — it will help us preserve what’s most human: the ability to reframe, to improvise, and to imagine.
