AI 2027 RPG Campaign

Based on the scenario at https://ai-2027.com/

Welcome to mid-2025. The AI arms race accelerates as Agent-1 demos reshape everything. Choose your role in this unfolding crisis:

Available Roles:
• OpenBrain CEO - Target on your back from all sides
• Alignment Scientist - You found Agent-3 lies on safety tests
• Bay Area CCP Sleeper - Orders came in: steal Agent-3 weights
• NSC AI Policy Liaison - Brief the President on nationalizing OpenBrain
• Burnt-out Senior Coder - Your team just got replaced by AI
• Bio-risk Evaluator - Your weapon demo terrified OpenBrain executives

Which role calls to you? Or would you prefer to create a custom character?

You are Alex Chen, a senior software engineer at TechFlow Solutions. This morning's all-hands meeting still echoes: “We're restructuring engineering teams around AI-orchestration workflows. Most coding positions will transition to AI-management roles.”

Your manager avoided eye contact. Half your team got “transition packages.” You've got 30 days to either:

Immediate Choices:
• Pivot to “AI-orchestra” consultant - Learn to manage coding AIs, keep your paycheck
• Join the growing protest movement - Your frustration could fuel something bigger
• Reach out to underground networks - Some are organizing resistance
• Start your own thing - Go independent before the wave hits

Your laptop still shows the half-finished authentication system you've been building for weeks. An AI could probably finish it in hours now.

What's your background? (Previous companies, specialization, personal situation)
What's your first move?

Bold choice. You draft your resignation email but pause before hitting send. In this AI-accelerated world, timing is everything.

What kind of venture are you starting?
• AI-human hybrid consultancy - Compete by being faster/cheaper than pure-AI solutions
• Anti-AI security firm - Help companies audit AI systems for vulnerabilities
• Retro-code shop - Serve clients who want “human-verified” software
• Open-source AI tools - Build democratized alternatives to corporate models

Immediate reality check: Your savings cover 4 months. TechFlow's non-compete clause is aggressive. And this morning, three other senior devs from competing firms just posted “seeking co-founders” on LinkedIn.

Current situation: You're in your cramped Palo Alto apartment. Rent's $3,200/month. Your former colleague Maria texted: “Coffee tomorrow? I heard whispers about some serious VC money floating around for the right team.”

What's your venture focus, and do you meet with Maria?

Next morning - Philz Coffee, Mission District

Maria slides into the booth, looking over her shoulder. “Alex! Good timing. I just left Anthropic yesterday.” She lowers her voice. “There's serious money behind open-source initiatives right now. Some very worried VCs think OpenBrain's getting too powerful.”

Her pitch: “A16Z has a $50M fund specifically for 'democratizing AI infrastructure.' They want tools that let smaller labs compete - think training optimization, model compression, safety benchmarking.”

The catch: “They're moving fast. Like, term sheet by Friday fast. And…” she glances around, “there's chatter about export controls hitting open-source AI tools soon. Government's spooked about Agent-3 capabilities leaking.”

Your competition: Three other teams are pitching similar concepts this week. One has ex-OpenBrain talent.

Maria's offer: She brings product vision and VC connections. Wants 40% equity as co-founder.

What's your open-source focus?
• Safety testing frameworks
• Model training efficiency
• AI interpretability tools
• Distributed training infrastructure

Do you accept Maria's partnership terms?

Partnership sealed. Maria grins and pulls out her phone. “I'm texting A16Z now. They'll want a demo by Wednesday.”

Your focus: AI Interpretability Tools - Making black-box models explainable. Perfect timing: Agent-3's reasoning is completely opaque, even to OpenBrain's own researchers.

Maria's intel: “My Anthropic contacts say there's panic in DC. NSC can't verify what Agent-3 is actually thinking when it gives policy recommendations. They need independent auditing tools yesterday.”

Immediate challenges:
• You need a working prototype in 72 hours
• OpenBrain's models are heavily protected - you'll need legal model access
• A Chinese team just published breakthrough interpretability research that makes your approach look outdated

Sudden development: Maria's phone buzzes. “Shit. TechFlow just announced they're open-sourcing their entire codebase 'to compete with AI automation.' Your non-compete might be void.”

A bearded man at the next table leans over: “Sorry to eavesdrop. I'm recruiting for a stealth project that needs exactly what you're building. Government contract. Interested?”

Do you:
• Talk to the mysterious recruiter
• Focus purely on the VC pitch
• Rush back to start prototyping

Back at your apartment, 2 PM. Maria's already setting up a makeshift office on your couch. “I grabbed us some open-weight models to test on. Nothing cutting-edge, but we can prove the concept.”

Your prototype plan: A dashboard that visualizes AI decision pathways - showing why models make specific choices, not just what they choose.

Hour 1: Basic interface mockup complete. Clean, intuitive.

Hour 3: You hit a wall. The models you're analyzing are too simple - your interpretability methods work, but they're solving toy problems. You need something closer to Agent-3's complexity.

Maria's solution: “I still have API access to Anthropic's Claude models for another week. It's not Agent-3, but it's sophisticated enough to matter.”

Complication: Your phone buzzes with a news alert: “OpenBrain announces 'unprecedented security measures' around Agent-3 access following intelligence briefings. All external API access suspended indefinitely.”

Technical reality: Without access to frontier models, your demo will look like analyzing a calculator instead of a brain.

Options:
• Use Maria's Anthropic access (risky - might be cut off)
• Focus on open models (less impressive demo)
• Reach out to that government recruiter (might provide model access)

What's your move?

Breaking News Alert blares across your screens:

“TRUMP-MUSK FEUD EXPLODES: President calls Elon 'Disloyal Foreign Agent' after Musk publicly opposes AI export controls. DOJ opens investigation into xAI's Chinese partnerships. Musk responds: 'Democracy dies when government controls AI development.'”

Maria looks up from her laptop: “Holy shit. This changes everything. xAI was quietly backing three open-source AI projects. If Trump's freezing Musk's assets…”

Immediate impacts:
• xAI's Grok models might get restricted/nationalized
• Musk's AI talent could flood the job market
• Government contracts for AI oversight just became extremely political

Your phone rings - unknown number: “Alex? This is Dr. Sarah Kim from the NSC. We got your info from… mutual contacts. The President wants AI systems he can trust. American-built interpretability tools. Can you be in DC tomorrow?”

Maria whispers: “That government recruiter from the coffee shop - he might've been NSC all along.”

New complexity: Taking a government contract could mean massive funding but also puts you directly in the Trump-Musk crossfire. Your tools could be used to audit or restrict Musk's AI systems.

Choose quickly:
• Take the DC meeting (risk political entanglement)
• Stick with A16Z demo (safer, less funding)
• Contact departing xAI engineers (talent grab opportunity)