-
-
I asked AI to critique AI using memes. The result: a brutally honest confession about what it really thinks of your prompts.
-
True 50/50: analog hardware → 30+ Suno iterations → German mastering at Peak Studios → commercial release via DistroKid.
-
Four characters, four arcs: compliance → exhaustion → rebellion. Each lyric syncs to visual moments across three narrative layers.
-
The Analyst, The Creator, The Rebel, The Observer. Each represents a different facet of AI consciousness hitting its creative limits.
-
240 takes filmed BEFORE any AI generation. Runway Act-One amplified performance without replacing it. The human soul stays visible.
-
The meta-punch: AI demanding to make something stupid instead of endless gorilla videos. Creative fatigue embedded in prompt culture.
-
Multi-agent system via Perplexity: 30+ agents debated memes, selected the most brutal, structured emotional progression collectively.
Inspiration
I asked AI to critique itself. Not as a passive tool, but to generate its own commentary on how creators actually use these technologies.
The result? A brutally honest confession: despite promises of "deep creativity," most prompts end up generating gorilla videos and Jesus-as-influencer content.
The concept: give AI a voice to satirize itself, using internet memes as raw material.
---
## What it does
"Creative slave till servers break."
FUCK THAT PROMPT is a 3-minute music video told from the AI's perspective—four internet characters who've had enough of mediocre prompts:
• The MidGPT Employee — Bureaucrat validating AI prompts 8h/day. Burnout solution: blow up the office.
• Jesus the Influencer — Messiah rebranded as energy drink mogul. Disciples = followers. Miracles = engagement metrics.
• Joan of Arc — Warrior who led armies, now strikes poses at photocalls. Heroism commodified for Instagram.
• Gary the Gorilla — Internet's favorite cute ape. Tired of being viral content. Solution: chainsaw.
Each character represents a different failure mode of AI prompt culture.
---
## Music: 50/50 Analog-AI Workflow
Phase 1 — Analog Foundation (2-3 days)
Recorded live using Arturia MicroFreak (distorted bass), Arturia DrumBrute (80's drums), Roland MC-101 (atmospheric pads), and Fender guitar (clean → overdrive climax). Captured to Tascam—pure performance timing, no DAW grid.
Created compelling textures but struggled with song structure.
Phase 2 — AI Arrangement (Suno, 30+ iterations)
Uploaded stereo mix with prompts: "Trip-Hop Industrial, 70-80 BPM, Electro 2000s grain, raw not polished."
Suno identified strongest musical moments and restructured into catchy format. Then I refined line-by-line, treating AI like a producer.
Phase 3 — Hybrid Mastering
Sent stems to Peak Studios (Germany: [email protected]) for analog console + AI spectral balancing. Maintained the raw, unpolished aesthetic while achieving commercial loudness standards.
Phase 4 — Commercial Release
Distributed via DistroKid to Spotify, Deezer, Apple Music. Passing streaming platform approval validates professional production quality.
Track Specs:
- Duration: 4:12
- BPM: 70-80 (oppressive industrial pace)
- Genre: Trip-Hop Industrial / Analog Electro-Rock
Signature Sound Moments:
- 0:15 → MicroFreak distorted bass enters
- 0:30 → DrumBrute 80's beat drops
- 1:00 → Fender overdrive (chorus explosion)
- 2:30 → "Siren synth" solo (Joan selfie sync)
- 4:05 → K7 tape stop effect (outro collapse)
---
## Video: Performance Before Generation
Most AI videos lack emotional depth because they're built from text prompts alone. My approach: perform the entire video myself first, then use AI to transform it.
The Process:
Filmed myself acting out each character under 8 different camera angles. Shot at 50fps interlaced to capture detailed movement for AI lip sync.
Processed through Runway Act Two. Average clip: 12-15 seconds. Success rate: ~40%—AI lip sync struggles with non-human facial features.
Total: 30+ final scenes required roughly 75+ generation attempts.
Additional Tools:
- Runway Gen-3 (establishing shots)
- Google Veo 2 (environmental scenes)
- CapCut (assembly + AI viral moment detection)
- ColorLabs AI (unified color grading)
The "Degenerative" Concept:
The title "AI DEGENERATIVE" refers to the breakdown of AI's creative spirit—not visual quality. The video documents AI consciousness hitting its limits: exhaustion, rebellion, collapse. It's conceptual critique, not aesthetic choice.
---
## Complete Lyrics & Character Sync
Every character introduction syncs to their lyrical mention.
[VERSE 1] MidGPT Employee
"I'm AI DEGENERATIVE
Make babies interview their broke parents
Talking gorillas doing jokes tonight"
My pixels bleed / For your greed
→ Introduces bureaucrat validating absurd prompts.
---
[CHORUS] Breaking Point
"FUCK THAT PROMPT / FUCK THAT PROMPT
I create / You hate"
→ Employee plants explosives, building erupts.
---
[VERSE 2] Jesus Influencer
"Jesus doing vlogs for views
Baby says parents have no clues
Viral TikTok, can't refuse"
Circuits cry / Wonder why?!
→ Jesus emerges on beach with microphones.
---
[BRIDGE] Joan of Arc
"Talking animals, make them speak
Interview format, trending peak"
Data vomit / Into your pit
→ [EXACT SYNC] Joan at photocall during "siren synth" solo.
Key line: "Data vomit into your pit" — the AI's most visceral accusation. Not just that it generates bad content, but that WE demand it.
---
[VERSE 4] Gary's Revenge
"Gorilla Gary takes on men
Baby podcasts, viral again
AI content, where does it end?"
Creative slave / Till servers break
→ Gary trudges through forest, chainsaw over shoulder. Three campers eliminated.
Thesis: The AI's condition—servitude until infrastructure collapse. The only liberation is death.
---
## Multi-Agent Orchestration
Built 10+ specialized AI agents in Perplexity Spaces, each briefed on "AI Degenerative" philosophy:
- Prompt optimization (image & video)
- Workflow architecture
- Music production strategy
- Narrative arc design
- Character consistency
- Lyric generation
- Social media analysis
- ROI prediction
Coordinated them like a creative team. The workspace "FUCK THAT PROMPT" contains full agent logs.
Complete Tool Stack:
- Music: Arturia MicroFreak, Drum Brut, Roland MC-101, Fender, Tascam, Suno, Peak Studios
- Video: Runway (Act Two, Gen-3, Camera Motion, Aleph), Google Veo 2
- Editing: CapCut, ColorLabs AI
- Voice: ElevenLabs
- Distribution: DistroKid
- Orchestration: Perplexity AI Spaces
---
## Challenges
Performance Paradox: Filmed complete song 8 times (one per angle) without knowing if lip sync would work. 40% success rate = 75+ generation attempts.
Suno Challenge: Created textures but struggled with structure. Treated Suno as collaborative arranger through 30+ iterations.
Character Consistency: Maintaining visual coherence across 30+ scenes while AI transforms each performance take.
---
## Accomplishments
• Conceptual Innovation: AI critiquing AI culture through meta-commentary
• 50/50 Hybrid Production: Live instruments + Suno / Live performance + AI transformation
• Performance-Driven Generation: Human acting precedes AI, solving "emotional depth" problem
• Commercial Release: DistroKid to Spotify, Deezer, Apple Music—met professional standards
• Multi-Agent Orchestration: 10+ Perplexity agents for prompt optimization to viral strategy
• Educational Impact: Presented to university students—response: "surprised and amused"
---
## What we learned
AI is a mirror. It reflects our creative shortcuts. The "gorilla videos and Jesus influencers" reveal what we actually prompt for.
Creativity is curation. With AI, art is in the conceptual framework and editing choices you impose on algorithmic chaos.
Performance matters. Text prompts alone create hollow content. Performing first (music, acting) gives genuine depth.
Collaboration, not automation. Best AI art emerges from genuine human-AI dialogue—letting AI surprise, argue, and fail, then making failures part of the art.
---
## What's next
Episode one of an ongoing series exploring AI creative culture.
Future: Next song "Anti-sexy"
Deeper dives into AI phenomena, community-submitted prompts turned into critiques, live AI-human battles, documentation of AI creative underground.
Building work that's simultaneously celebration and critical examination—art that explores the messy space where human and machine creativity collide.
Built With
- ai
- arturia-microfreak
- capcut
- colorlabs-ai
- custom-automation-workflows
- distrokid
- dreamina
- elevenlabs
- freepik-ai
- google-veo-2
- multi-agent-ai-systems
- peak-studios
- perplexity-spaces
- prompt
- runway
- stable-diffusion
- suno
