Category	Definition	Example(s)
Fake	Simplified models good enough for specific purposes	AI
Imaginary	Something we can never create1	AGI, ASI
Real	Exists	Human minds

Training a model is not teaching a model. The “alignment” problem is not fixable. As a computational physicist with a PhD in ML, I know every technical “AI breakthrough” has been about improved feigning of intelligence, not actually being intelligent. One tiny subcomponent of the problem is jaggedness, which is hilariously catastrophic, while hallucination is mathematically unfixable. ML is founded on compressed sensing, which we often joke is “glorified statistics”. I think misplaced glorification is no joke.

---

1. In the story of the Tower of Babel, though humans might have the raw power and technical ability to build anything, their endeavours fail regardless – because the flavour of ambition required ironically causes people to stop building and start fighting. ↩