The dead internet

Imagine this: IDC’s Global DataSphere forecast reported 64.2 zettabytes of data created or replicated in 2020 and projected 23% annual growth through 2025. That works out to roughly 120 zettabytes in 2023 — or 120,000,000,000,000 gigabytes in a single year.

AI-generated material is adding to that torrent, but there is no reliable figure for what share of all new content it represents.

Bots and AI systems already churn out vast numbers of texts, images, and videos every day.

You think you’re liking your friends' posts or reading the opinions of real people — but you’re actually talking to algorithms.

The pendulum is already swinging the other way: standards such as C2PA Content Credentials can attach tamper-evident provenance information to media. They do not prove that something was made by a human — or that the persona behind it is human — but they can help show where a file came from and how it changed.

Provenance can help with the file, but it cannot settle the social question: when an avatar speaks in your voice, is it still “you”? Would you create one?