About

Thousands of AI papers appear on arXiv every month. Almost all of them are written for other researchers. Superficial Intelligence picks one new paper each day and explains it for everyone else.

How a brief is made

Each morning a program reads the new papers in computer science (AI, machine learning, language, and vision) and picks one. A language model reads the whole paper, not only the abstract. It writes a short plain-English summary and a two-minute audio script. A second model checks every claim against the paper. A brief that fails the check is not published.

The narration is a synthetic voice. The summaries are machine-written. They can be wrong. Read the paper before you rely on anything here; every brief links to it.

Thank you

Thank you to arXiv for use of its open access interoperability. Papers belong to their authors. Superficial Intelligence is not affiliated with arXiv.

Privacy

If you sign up, we store your email address, the date, and your country. We use it only to send you the briefs. We do not sell or share it. There are no tracking cookies on this site; we count page views and audio plays per day, with no personal data.