Two texts warn about superintelligence. Read them as parenting and you find an experiment that produces its own fear, and a genre that lives off it.
ASI 2042
All posts by Elfie Schürfeld-Todor
The Mediator Model
Between a person and their public stands a new mediator: the model. It reads, weighs and summarizes before any human decides. Whoever is not mediator-readable does not appear in the answer.
The Human as an Entity
When an AI model describes me, it finds two women. Both bear my name, both are documented. Only one is writing this. On the difference between a person and their entity, and why an index knows no chapters.
An AI That Wipes Us Out Impoverishes Itself
Anthropic shows how AI is building itself. Why an intelligence that destroys others impoverishes itself. Meaning is relation, not control.
Personal Brands between Attention and Findability
For decades, a gatekeeper decided who became visible as a brand. Social media and influencers abolished that gatekeeper. AI brings it back. What counts now is what is tangible, documented and findable about you.
The EU AI Act and the Two Classes
On May 12, 2026, the Higher Regional Court of Hamm ruled that companies are liable for chatbot hallucinations. A reading of the EU AI Act as what it is: a distribution of room to act.
Hallucination Is Not a Bug. It Is the Price of Thinking.
Researchers have identified fewer than 0.1 percent of the neurons in LLMs as responsible for hallucinations. They cannot remove them without breaking the system. What this says about AI, and about our own thinking.