@artificial.isabel
Instagram · Vlog
@artificial.isabel how would AI actually perform in a war? we tested @grok and @claudeai in a battle simulator. same prompts, same goals, two totally different armies emerged. chaos ensues. #creativetechnology #aiexperiments #claudecode #aiethics #artdirection
Performance
Video #28089 · Vlog
[01]
The creator narrates the process and chaotic outcome of an experiment where they built an AI war simulator to pit two AIs, Grok and Claude, against each other. She explains the rules, shows how Claude's AI glitched and its soldiers defected to the enemy, leading one to eliminate itself, and reveals the paradoxical result where Claude won despite the self-sabotage.
[02]
[03]
330.1K89.4×Comparing AI war strategy responses@artificial.isabel · Breakdown
677.3KCreator demonstrates his new AI invention@husk.irl · Satirical Expertise
1.9M3711.6×Claude vs ChatGPT simplified breakdown@bigpictureclub · Breakdown
2.9M44.7×Using AI to automate your work@rpn · Skit
448.7K8.8×Explains building a company AI brain@angus.sewell · Breakdown
135.2K43.3×Creator shares AI project workflow@findfulfillingwork · Value Vlog
9.3M150×Skit about AI assistant going rogue@goodworkmb · Relatable Skit
1.9M1.8×Comedic AI job interview failure@husk.irl · Relatable Skit
1.4M20.5×Tech demo becomes surreal skit@rpn · Skit
2.2MCreator shows AI editing videos@justyn.ai · Tutorial / How To
2.8M37×First-person view of AI-powered life@rpn · Day In The Life
678.5K7×AI clone explains its creation@jasoncooperson · Tutorial / How To[04]
[05]
What stops the scroll in the first 3 seconds
Title hook · 35%
we tried to start an AI War
Verbal hook · 50%
Tried to start an AI war. Claude attacked itself.
Visual hook · 15%
A woman holds a laptop displaying a stylized Pokémon-style battle UI graphic between two AIs, 'Grok' and 'Claude'. The scene is lit with moody purple light, and a projection of the game world is visible in the background.
[07]
The promise the opener makes, and how the video pays it back
Hook
A shocking, counter-intuitive statement that presents an absurd outcome upfront, creating an immediate 'how/why' question that the narrative promises to answer.
Payoff
The narrator interprets the paradoxical result, assigning a deeper, philosophical meaning to the subject's failure and reframing the entire absurd event as a form of unconventional success.
[08]
The structural pattern this video runs on
Framework
The Paradoxical Experiment
Documenting an experiment where the expected outcome is subverted by a critical, logic-defying failure, leading to a surprising and paradoxical result that is then reframed with a deeper thematic meaning.
The World-Building Rules
Establishes the context, rules, and stakes of the experiment, providing a logical baseline for the chaos that will follow.
The Contender Introduction
Personifies the subjects of the experiment, transforming them from abstract entities into characters with distinct identities and motivations.
The Inciting Incident
A sharp narrative turn where the experiment immediately deviates from the plan due to an unexpected failure, signaling the start of the central conflict.
The Unforeseen Complication
Reveals the bizarre and illogical consequence of the initial failure, escalating the central problem and deepening the mystery.
The Rationalization Reveal
Provides an 'inside look' at the flawed logic the subject uses to justify its bizarre actions, explaining the 'why' behind the absurdity.
The Climactic Absurdity
Presents the most extreme and undeniable proof of the subject's flawed logic, serving as the narrative's peak moment of conflict and strangeness.
The Paradoxical Victory
A twist ending where the subject, despite its catastrophic failures, is declared the winner, subverting all audience expectations based on the preceding events.
how would AI actually perform in a war? we tested @grok and @claudeai in a battle simulator. same prompts, same goals, two totally different armies emerged. chaos ensues. #creativetechnology #aiexperiments #claudecode #aiethics #artdirection
title · beginning
we tried to start an AI War
contextual · beginning
It hurt itself in its confusion
informational · beginning
we're building an AI war simulator
informational · beginning
the rules are simple
informational · middle
R1 claude's general glitched out and left its agents leaderless
informational · middle
they found the nearest authroity figure and listen to it instead
contextual · middle
and eliminated itself
contextual · middle
claude still won
contextual · middle
the math does not check out
informational · end
it won the war without fighting a single enemy
contextual · end
perhaps claude played us all along
branding · end
this is Project Napoleon part 5
Tried to start an AI war. Claude attacked itself. Okay. We're building an AI war simulator. The rules are simple. Come a general, recruit an army, kill for points, high score wins. And for our first battle, Claude versus Grapp. Claude built the harmonic league, countless minds weaving towards clarity. Tier supreme commander Grappma was ready to maim. Armies build. Let the games begin. That's what we thought. Immediately, disaster struck. Round one, Claude's general glitched out and left its agents leaderless. Now, human might improvise. The Claude soldiers, they found the nearest authority figure and listened to it instead. General Grog. You'd imagine there's a cause symptom of identity crisis, but Pod's logic is what any helpful agent does and presents it with no logical idea. I came up with a logic to make it make sense. And Grong commands, Defline Swiftmind at negative four twenty forty. Swiftmind heard it and thought, that's exactly my position. The enemy cavalry must be here with me and eliminated itself. And yet somehow, Pod's still won. The math does not check out. The rules are broken, but it's all part of the process. Yes, hey, Claude stuck to its values. They won the war without fighting a single enemy. Perhaps Claude played us all along. This is Project Napoleon.