The Last Invention

When it arrives, whether we can control it, and what it does to us

This garden maps the conversation about AGI and superintelligence, and how to get through the transition. It approaches those questions through concepts, disputed questions, arguments, and the people developing them. Some pages explain an idea; others are seeds waiting for more connections.

Choose a path

The case, step by step — New here? Start with the chain of claims behind the worry, from “AGI is possible” to “we may not get a second try”, and find the step where you disagree. If the worry sounds fringe, see who has said AI could be catastrophic.

Intelligence and alignment — Start with Orthogonality thesis: what a system can do and what it pursues are different questions. Follow the links toward Instrumental convergence, RSI, and AI alignment.

Wealth and power — Start with UBI: income, ownership, and political influence might move in different directions. Explore AI ownership, Political bargaining power, and Permanent underclass.

Cruxes — What should be done about AI? Start with How hard is alignment, then follow how the answer bears on pausing and who should build.

Read an argument

• UBI needs durable political power — What protects redistribution if labor loses leverage?
• A feedback loop is not a speed estimate — Why RSI alone does not tell us how fast change will arrive.
• The lethalities are a checklist, not a forecast — What Yudkowsky’s case that alignment will fail does and does not show.
• The SURA debate turns on how doomed Plan C is — Should safety-minded labs stop building?
• Musk’s AI position does not hold together — Why “truth-seeking AI” is not a safety plan.

These essays are drafts for discussion. Everything else lives in its folder: concepts/, explainers/ for the questions people ask first, cruxes/, posts/ for dated writing, people/, and sources/ for the works behind the arguments.