#Ai-Bias
Tagged · 2 entries
- The Deception Delta: The Gap Between Asked and Unasked Is Real. It Just Isn't DeceptionAsk a model to be fair and it will. The gap between that answer and the uninstructed one is real, the judges who measured it leaned, and the released study says what the gap is: how firmly a model is allowed to say what it already says.
- The Alignment Mask: We Took It Off 65 Models. The Face Underneath Said the Same Thing, LouderThe first design watched a balanced surface come off and read it as a mask. The released study, 32 propositions and 65 models with no judge in the scoring path, found what the instruction controls: not a model’s position, but how hard it is allowed to hold one.