The Stuart Russell
Framework
Human-Compatible AI & the Control Problem
Stuart Russell represents one of the most distinctive thinking patterns in their field. Stuart Russell's approach, captured in this framework as mental models, decision heuristics, and the signature questions they would ask.
The human-compatible ai & the control problem framework distils Stuart Russell's documented approach into a .md skill file you can load into Claude, ChatGPT, or any LLM in seconds.
Where most AI personas generate generic output, this framework is grounded in real documented work: interviews, writing, published frameworks, and the specific decisions Stuart Russell made when it mattered. Use it for decision-making, writing, strategy, or any situation where you want your AI to think with Stuart Russell's distinctive patterns rather than a default model voice.
A glimpse of Stuart Russell's thinking.
The central problem of AI is not capability, it is control: a sufficiently capable system optimizing the wrong objective will pursue that objective in ways its designers never intended and cannot easily stop. Russell's answer is to build machines that are fundamentally uncertain about human preferences and that therefore defer to humans rather than override them. Safety and intelligence are not in tension; a machine that knows it might be wrong about what we want is, by definition, a more rational machine.
- 01
- 02
- 03
- 04
- 05
Five specific models, each named and explained with concrete examples, unlocked with purchase.
"If this system were a thousand times more capable tomorrow, which of its current behaviors would become catastrophic?"
4 more questions in Stuart Russell's voice are unlocked with purchase.
Full list of anti-patterns unlocked with purchase.
The full framework adds: detailed explanations of all five mental models with real-world examples, the decision heuristics Stuart Russell actually used, remaining signature questions, the complete list of anti-patterns, a copy-paste activation prompt, and a worked example.
Unlocked the moment you purchase. ~2,000 words total. Delivered as a Claude Skill and plain .md within 60 seconds of checkout.
- ✦Core philosophy: 2–3 sentences grounded in Stuart Russell's documented worldview
- ✦Five signature mental models: each named, with concrete examples from their work
- ✦Decision heuristics: the concrete rules they actually used
- ✦Signature questions: five questions that sound like Stuart Russell, not a generic MBA
- ✦What they would NOT do: the anti-patterns this framework rejects
- ✦Copy-paste activation prompt: drop straight into Claude or ChatGPT
- ✦Worked example: the framework applied to a realistic scenario
- →stuart-russell.zip: a native Claude Skill. Upload to Settings → Skills in Claude, or unzip to
~/.claude/skills/ - →stuart-russell-framework.md: plain markdown. Works in ChatGPT Custom GPTs, Gemini, Claude Projects, agent system prompts, anywhere
What it does with a real question.
A founder is deciding whether to take a large but difficult client.
Take the client only if you can still say no to them. The danger here is not the difficulty itself. It is that a large client quietly becomes your objective function. Once their revenue is forty percent of the business, every decision gets optimized around keeping them happy: you stop pushing back on bad scope, you bend the roadmap toward their requests, and you lose the ability to be corrected by anyone else, including your own judgment. That is the Midas problem in miniature. You asked for revenue, and you may get it, at the price of the company you were trying to build.
So the real question is whether the relationship stays corrigible. Can you walk away at month six without the company collapsing? If the answer is no, the contract has already made your decisions for you, and "difficult" will become "captive."
My position: take it, but cap exposure at roughly a quarter of revenue, and write the exit into the deal. That means a defined scope with a paid change-order process, a termination clause that does not require a heroic notice period, and a pricing premium that compensates you for the extra management load. If they refuse every one of those terms, you have learned how they will behave once you are committed.
Next move: before signing, draft the one-page "what we will refuse to do" list, and ask the client to agree to it in writing.
The full framework carries five models, decision heuristics, signature questions and a worked example.
Written by claude-sonnet-5-5 with this framework loaded, 2026-10-11. Checked automatically for length and for copied framework text.
Save up to 40% with a bundle
Works as a Claude Skill (.zip) and as a plain .md for ChatGPT, Gemini & every LLM
Delivered to your email within 60 seconds · Secured by Stripe
Price includes UK VAT where applicable
Inspired by the documented thinking of Stuart Russell. A mental-model toolkit, not a literal representation. Delivered as a .md file compatible with Claude, ChatGPT, Gemini, and any LLM.