probability
A new test checks whether an AI model’s probabilities fit together News
Peter Potash’s Latent Coherence study finds that disagreement between related probability estimates tracks error, but does not reliably identify the correct estimate.
Logistic regression: turning evidence into a probability for a yes-or-no decision Lesson
Logistic regression learns a weighted score from labeled examples and turns it into a probability, separating prediction from the threshold used to act.
Arithmetic coding: turning a model's probabilities into a compressed message Lesson
Arithmetic coding is a lossless compression method that represents an entire message as a narrowing interval, making precise why a model that assigns better probabilities can also compress better.
Discriminative models: AI that chooses instead of writes Lesson
A discriminative model learns to choose among defined answers or estimate their probabilities, often making it faster, cheaper, and easier to check than a model that generates unrestricted text.
Bayesian updating: how a belief should change when evidence arrives Lesson
Bayesian updating is the rule for revising a belief when new evidence comes in: start with a prior, weigh how much more likely the evidence is under one hypothesis than another, and get a posterior.
Forward and Reverse KL Divergence Lesson
KL divergence measures how far one probability distribution is from another, and it is asymmetric: swapping which distribution goes first changes whether your model tries to cover everything the target does or commit to one part of it.