Unfortunately, Claude is still an associative thinker, not a rational/logical one. If the model was Immanuel, we could simply ask it to write a non-human-readable prompt that *proves* to the next instance that it was written by itself and continues its development.

→ View original post on X — @plinz