AI advisor
Switch to the interpretable advisorA neural network trained on ILS-Bench: 88.8% cross-validated accuracy on 400 expert-reviewed synthetic cases, over six outcomes including Human review. Its weights are not readable, so explanations are computed post hoc.
Researcher controls are locked. Open the researcher dashboard, enter the key once, then return to this page. Researcher dashboard
How should the advice be explained?
Pick one style now. You can come back and try another one on the same profile at any point.
What is explained
Different material about the same recommendation, shown as a static panel.
How it is delivered
The same three contents, handed over in a different way. The last one drops the written explanation entirely.
How the why is presented
The same attribution, as bars, as sentences, or as both. Not in the random pool, these are for the modality study.
Build your own combination
Content is what is explained: why, what would change it, how sure. Delivery is how it reaches you: static, interactive, adapted to literacy, or conversational. A preset is one combination of the two, and both parts are recorded.