AI & Scientific Discovery
Start with a concrete question and distinguish a primary paper from marketing commentary. Record its date, task, data and measurement. Comparing models requires compatible conditions rather than two numbers taken from different tables.
Trace claims to their origin
Start with a concrete question and distinguish a primary paper from marketing commentary. Record its date, task, data and measurement. Comparing models requires compatible conditions rather than two numbers taken from different tables.
Keep the original link and identify what applies to your work. Attribute untested findings to their source instead of presenting them as personal experience. Distinguish a released feature from a research demonstration.
Follow the evidence chain
For a claimed improvement, identify the original paper, its comparison and its evaluation set. Check whether the result is measured by people, an automated score or a selected demonstration. Preserve that distinction when translating a technical claim into a short article.
Draft a qualified caption
Write a caption containing the task, test conditions and source. Then remove any broader conclusion the source did not measure. If a demonstration is a reconstruction made for your article, label it as such and keep it separate from figures supplied by the researchers.
