CONCEPT
The Is-Ought Gap
David Hume’s 1739 observation that no description of how things are can logically entail a conclusion about how they ought to be—a gap that no amount of training data, behavioral imitation, or preference learning can bridge, and that makes the specification of machine values an irreducibly human act of choice rather than a technical derivation.
In a paragraph so brief it has the texture of an aside,
David Hume noticed something in 1739 that remained unresolved for three centuries and has now become the deepest structural problem in AI alignment. Writers on morality, he observed, begin by making claims about what
is the case and then suddenly shift to claims about what
ought to be, as if the second follows from the first without explanation. He pointed out that it never does. A valid argument cannot contain in its conclusion what is not present, at least implicitly, in its premises. If every premise is a statement of fact, the conclusion can only be a further statement of fact; there is no rule of inference that takes you from description to prescription, from “this is how the world is” to “this is how the world should