Thin content is any page that doesn't offer enough genuine value to justify its own place in search results, whether because it's too short, too generic, duplicated from elsewhere, or built without a real reader's need in mind. Google's own quality guidelines name several categories directly: automatically generated content with no human oversight, doorway pages built only to funnel traffic, scraped or lightly reworded content taken from other sources, and thin affiliate pages that add nothing beyond a reworded product description and a link.
Thin content doesn't just underperform on its own page; a site with enough of it can drag down how Google evaluates the site's overall quality, which is part of why a large corpus of weak pages sometimes correlates with weaker performance across pages that are actually good. Fixing it depends on the page: genuinely useful content underneath weak execution can be expanded and deepened, a page that duplicates something better elsewhere on the site is better consolidated via a redirect or canonical tag, and a page with no real reason to exist is usually better removed outright than left live at a low bar.
