The AI spammers are coming

If your media diet is at all similar to mine, it’s likely your Twitter (err, Mastodon) feeds are filled with screencaps of disturbingly well-written, sometimes-correct output from Open AI chat. Generative AI with tools like GPT-3 raises the potential for massive disruption of many industries, with its ability to produce deeply convincing content incredibly cheaply. While the opportunity for positive impacts is significant, the challenges it poses to the media industry are existential.

Below is a synthetic obituary written for the very-much-alive Glenn Danzig.

It’s well-written, convincing, and entirely incorrect. Producing disinformation like this has suddenly become vastly cheaper than it was just a few months ago, and with no marginal cost to produce this content we should expect a massive spike in its (already high!) production. Trustworthy journalism that can find a way to drown out the noise has never been more important, or more challenging.

But wildly cheap-to-produce disinformation isn’t the only threat the media faces. Wildly cheap correct information has the potential of being just as disruptive. For example, the below article could certainly rival the content produced by many SEO-focused sites.

Will there be any place at all for human-written SEO-friendly content? It seems likely that the market for it will be massively smaller, as it struggles for position in search results against a broad set of algorithmically generated content.

Other forces acting on the media industry, like the disintermediation from audience by platforms, have resulted in slow, yet significant shifts over the past decades. By contrast, the change in online content driven by autogenerated media will be just as large, and not at all slow — we should expect many parts of the media world to look quite different in 12 months than they do today.