Where engineering meets philosophy

If a model confidently gives you an answer you did not know was wrong, what did your brain actually learn?

• Epistemic dependency.

  Aligning AI is also a human problem, not only a technical one, and the misalignment is already happening in offices.

• Bias at work.

  The most dangerous AI is not a super-intelligence. It is a mediocre model that sounds right most of the time and is used without oversight.

• Ethics in practice.

  Models produce something that resembles understanding without any lived experience behind the words. That deserves care.

• Artificial and natural reasoning.

  We trust authority because it sounds sure. We now do the same with confident machines, and that should worry us.

• Working together.

  I want AI to extend my thinking, not replace it. The day I stop noticing its mistakes is the day I stop being the engineer in the room.