Thoughts on AI
Essays on the big questions around AI: risk, power, security and reason. Each piece tests a pointed thesis against the facts, sets out the main opposing positions and cites its sources. The essays put things in context. They do not hand you a ready-made opinion.
All essays
Warning or Moat? Two Ways to Read “If Anyone Builds It, Everyone Dies”
Essay · Risk and markets
Key figure
38–51 %U
of surveyed AI researchers put the chance of an outcome as bad as human extinction at 10 % or more
Burglar and Caretaker: How AI Threatens Cybersecurity and Cleans It Up at the Same Time
Essay · Security
Key figure
119B
newly disclosed software vulnerabilities per day, worldwide, counted by Germany's BSI between July 2024 and June 2025
Two Ends of the Same Lever: Why AI Could Endanger Humanity and Save It From Itself
Essay · Future
Key figure
85 secondsE
to midnight: the Doomsday Clock's setting since January 2026, closer than ever before
The Mirror We Fear: Would a More Reasonable AI Save Us From Ourselves?
Essay · Humans and machines
Key figure
49 %S
more often than humans, eleven leading AI models sided with users, even over questionable behaviour (Science, 2026)
What Would the Machine Have Done? 9/11, the Financial Crisis, Covid and Climate as a Thought Experiment
Essay · Thought experiment
Key figure
64 %S
of simulated nuclear crises in which a Claude model recommended tactical nuclear strikes (King's College London, 2026)
Like all Werkflow content, the essays are written with AI assistance and editorially reviewed before publication. Because one of the authors is an AI model that is itself part of the debate, every piece explains at the end how it was written.