I Tried to Make AI Writing Sound Human by Banning AI Words Through logit_bias
I tried to make AI writing sound more human with logit_bias, an API setting that changes how likely a model is to…
Most searches now end without a single click to an actual website. AI companies are training their models on publisher content while destroying the traffic that funded creating that content in the first place. It's a pretty obvious value extraction problem.
Meanwhile, journalism jobs are vanishing. Major news organizations keep cutting staff. Entire regions have lost their local news coverage completely.
AI models get worse when they're trained mainly on AI-generated content. They lose accuracy and start making more mistakes. So AI genuinely needs fresh human-created content.
AI companies might pay licensing fees to publishers. But that doesn't come close to replacing what publishers lose when traffic drops. When someone visits your website, they might subscribe, come back repeatedly, buy things, attend your events. A one-time licensing payment doesn't capture any of that long-term value.
That expensive content that AI can't create on its own is exactly the content that's losing its funding. Investigative journalism takes months of work and significant resources. That's the stuff AI can't replicate.
This paradox needs a solution. Filling the internet with low-quality content doesn't help anyone, not even the AI companies whose future depends on the very content they're currently destroying the economics for.
Give Vroni a GitHub issue, bug report, spec, or rough idea. It reads the repo, plans the change, writes code, runs checks, and works toward a review-ready pull request.
Take a look at vroni.com