AI generates a lot of junk

AI poisons the information well

Together with

The Future. As AI-generated content becomes more ubiquitous online, more useless garbage is filling up the web, making it harder for people to sift through what’s helpful, relevant information and what exists simply to garner clicks and ad revenue. And as that content spreads, it may ironically infect the large language models that use the open internet to source information — a phenomenon known as “model collapse.”

Internet overload
The internet is getting inundated with what WSJ calls “a new kind of spam.”

  • Publishers are getting sent a flood of AI-generated articles far below their standards, prompting some outlets to close their portals to third-party content.
  • NewsGuard found 277 fake news websites pumping AI-generated content to game Google SEO and attract ad dollars.
  • That’s led to a cottage industry of how-to videos for using AI-generated content to make money — no matter how erroneous that content may be.
  • And more concerning, some are using AI to create fake phishing websites that can be used for scams.

Luckily, AI-generated content is easy to spot. According to Neil Clarke, publisher of sci-fi magazine Clarkesworld (an outlet that had to stop accepting submissions temporarily), it typically has “perfect spelling and grammar, but a completely incoherent story.”

We guess even the Terminator has problems with the three-act structure.


E-bike glow-up 🚴‍♂️💨

Want a reconditioned e-bike? Who doesn’t? Take Upway’s quick three-question quiz and find your best bike match. And they’ll zip it to your door in two to five days. 📦

Time to e-ride