In a joint op-ed published earlier this month by Maya Schenwar, board president for Truthout and director of the Truthout Center for Grassroots Journalism; Negin Owliaei, Truthout’s editor-in-chief; and Ziggy West Jeffery, Truthout’s executive director, the trio noted that Google’s AI model, Gemini, is now prioritizing summaries of news articles at the top of its searches, leading to dramatic drop-offs in viewership for independent media websites.
“We lost 15 percent of our readership overnight,” the op-ed writers said. “This was a readership we painstakingly built over decades, and it vanished because Google chose corporate greed over a healthy democracy.”
Independent media like Truthout largely rely on donations from readers, who are asked to consider donating when they visit these sites. However, when Gemini summarizes these articles (or prioritizes corporate media instead in its summaries), the loss in revenue is noticeable.
“At Truthout, we anticipate a loss of roughly $500,000 in online revenue as a result of this massive traffic plunge in the next year alone,” the op-ed said.
Bit of a tangent, but has anyone noticed AI is getting worse?
I use it for cvs and the like, because most companies also use AI to filter, but I’ve noticed it makes more mistakes than it did not even that long ago.
This was inevitable, the value of AI is derived from quality open and accessible training data of genuine expert human conversations that are constantly added to and kept up to date and yet the fundamental business model of AI companies is founded on disguising the value of that with a purposefully obscure woodchipper that brutalizes all of that information into a less useful, more manipulatable product of lossy compression algorithms.
The only way these AI companies can succeed is by definition completely devaluing the datasets they existentially rely on to maintain and create the AI, the only way this could ever go was to have a brief honeymoon period of AI companies stripmining the collective human wealth of the internet and then drowning and killing it with worthless bot crap that is always somehow fed back into the mouths of the very same algorithms creating the crap and thus everything degrades.
The only play these companies really have is to enshittify all other alternatives faster than their AI goes to shit… which is exactly what these companies are trying to do.
In the end all AI companies have really proved is that intelligence begins somewhere in the process of a librarian creating a library, and that we can gesture at that crudely by mutilating libraries created from the labor and love of countless people into untrustable bots that serve to privatize the wreckage of that knowledge… but none of that actually proves what the AI comapanies are doing is the valuable part it just proves that a well curated dataset contains a latent capacity for intelligence already.