prompt-atlas-ecl

Chapter 3 · The AI Aesthetics Frontier — Expansion

Canonical: PROMPT_ATLAS.md#ch3-ai-aesthetics-frontier · Prompts: prompts/ch03.yaml · Part: II

Worked Example — Adaptive Murals

Original prompt: “Design public art that responds in real time to weather, air quality, and human voices.”

1. Choose a wall + sensors — A public façade wired to temperature, PM2.5, and an opt-in microphone array (with on-device anonymization, no audio leaves the panel).
2. Define the visual grammar — A small generative model running locally with a style anchor contributed by a community artist; sensor inputs perturb hue, density, motion — never the anchor.
3. Provenance plate — A QR plaque names the human anchor artist, the sensor authority, and the model — read in any language.
4. Veto channel — A community council can pause, alter, or retire the mural at any time; the council’s seat allocation is published.
5. Sunset rule — At year five the mural is archived (not deleted) and the wall returns to the community.

Prompt Templates

# Co-author a piece
"Design a  co-authored by
 , an AI trained on , and
 .
 Specify the provenance plate, the veto channel, and the sunset rule."

# Aesthetics audit
"For , list (1) whose voices are amplified, (2) whose are erased,
 (3) the smallest change that would re-balance the mix without losing the
 anchor artist's intent."

Anti-patterns

• Style strip-mining. Using a model to generate “in the style of” a living artist without consent or compensation.
• Algorithmic uniformity. Recommendation-driven aesthetics flatten the world even as they multiply images.
• Beauty as bait. Awe weaponized for engagement metrics is not awe; it is dependency.
• Silent provenance. Any AI-augmented piece without a clear plate of co-authors is laundering authorship.

Try This

1. Provenance Plate — Write the plate for one piece you have made (or might make) with AI assistance.
2. Canon Audit — List the dominant cultural canons in your prompt’s training set; name three you are not drawing from.
3. Living Mural — Sketch a one-week trial with a sensor and a local artist.
4. Veto Channel — Design the smallest possible community veto for an AI-augmented public artwork.
5. Sunset Rule — Decide, in advance, when your AI-augmented piece will be archived.

Repo Cross-Links

• MACP bus — sensor/agent topics for the mural example.
• CoLearningMemoryStore — long-form archives of community-contributed art (with provenance metadata).
• server/core_bridge.py — pattern for an opt-in memory toggle (PAE_MEMORY=1).

Guide for AI & Humanity

• Consent before canon. Living artists must opt in. Dead artists deserve named attribution.
• Local before global. Aesthetics rooted in a specific place beat aesthetics scaled by API.
• Sensors are political. A microphone in a plaza is a relationship; design it as one.
• Reversibility. Every algorithmic art piece in public space needs an off switch and a sunset clause.

Citations & Further Reading

• Ursula K. Le Guin, The Carrier Bag Theory of Fiction (1986).
• Lewis Hyde, The Gift (1983) — the difference between commodity art and gift art.
• Yuk Hui, Art and Cosmotechnics (2021).

This site is open source. Improve this page.