Imagine you could interview thousands of educated individuals from 1913—readers of newspapers, novels, and political treatises—about their views on peace, progress, gender roles, or empire. Not just survey them with preset questions, but engage in open-ended dialogue, probe their assumptions, and explore the boundaries of thought in that moment. This is what time-locked language models make possible. Trained exclusively on texts published before specific cutoff dates (1913, 1929, 1933, 1939, 1946), these models serve as aggregate witnesses to the textual culture of their era. They cannot access information from after their cutoff date because that information literally does not exist in their training data.

Read more.

Posted in

Leave a comment