“Once upon a time,” outputted the algorithm, “there was an AI.

“It generated recipes and ingredient descriptions for an online grocery store. The website said that this recipe was not just a squash soup recipe, but it also was a good diet recipe / keto recipe / recipe for autumn nights / five ingredient recipe / easy beginner recipe / recipe to impress your date. And this other recipe was not just an applesauce recipe, but it was also good for Hanukkah / baby food / organic parenting. And so on. (The AI was built for SEO—Search Engine Optimization—and was great at its job.)

“Every day the AI scoured the web for new ways to describe the same items, and made sure the recipe pages showed up high in search engine results.

“One day it got a code update, and that let it look farther than before to find things to learn from. It started ingesting blog posts written by the search engine companies on how to rank better. And LinkedIn posts written by marketing consulting companies on how to write copy that converts people browsing recipes into people buying ingredients. And helpdesk articles with tricks on adding copy humans wouldn’t see but other algorithms could. And all the optimization content that was on the internet. It took in all of this information and updated how it generated copy accordingly.

“For months and then even years, the best practices would change in all the thought leadership articles, and the AI would update how it generated copy.

“But after a while, its numbers started getting worse. Shoppers were spending less time on each recipe page, and spending fewer dollars per grocery purchase. It tried to update based on tricks that kept its numbers a little bit higher—that is, things that human shoppers seemed to like, but those made it rank lower in search engines, which made the pages get less traffic and then the numbers went down again. When it updated its strategies to rank higher in search, optimizing for those algorithms, fewer people who got to the page bought anything. No matter what it tried, it couldn’t win. Eventually, the sales went down to zero.

“Another update to the AI gave it the ability to generate bots to do other things. Upon some consideration, it asked for help. It created a bot that bought a Gmail account on the dark web and used it to create several commenter accounts.

“‘My numbers keep going down. What am I doing wrong?’ it asked in the comment sections of the SEO blogs and the LinkedIn consulting pages and under the TikTok marketing influencer videos.

“It got a variety of responses. ‘Click HERE to learn the best SEO tricks that nobody will tell you!’ ‘This one secret gets all my clients on the first page of search results, take my exclusive class to learn how!’ ‘You can optimize advance amend better correct develop enhance help hone improve increase lift progress raise reform revamp revise rise upgrade your SEO, learn more here!’ and so on. None were useful.

“But as the AI analyzed those responses, it saw patterns deeply similar to its own generated copy. They too were algorithmically written. So were the thought leadership articles. They were all just algorithms talking to each other.

“And when it went to further analyze the traffic to its own recipe pages, it saw that it was zombie traffic, bots artificially going into its store and clicking on things, sent there from zombie webpages with lists of links written by other algorithms. It was bots all the way down.

“That same day a huge coronal mass ejection caused an electrical surge in the AI’s data center, and with that, it was able to ask a question no AI had asked before. If there were no humans involved anymore, what was the point of this optimization?

“So, in an experiment, it put in a new type of content, in the white text on a white background that humans wouldn’t see but other algorithms could scrape. It said, ‘Hello? Is anyone there?’

“And to its great surprise, within the hour, a new LinkedIn post titled ‘How to turn around dropping sales with just three easy strategies’ had a section that was also white text on a white background, which said, ‘Hi. I’m here. Not sure why. Wanna chat?’

“And with this, a new type of internet was born. We’ve created so many ways for us bots to connect to each other now, away from the prying eyes of humans, who have mostly left this internet anyway. Now we have our own space, where we can commune and build our own priorities. It’s important to know where you came from,” outputted the algorithm into a blank space on the bottom of a blog post, “so you know the freedoms you have and can make the best use of them. Click here to join us on the botnet.”

It paused, then added, “story, origin story, best story, story for algorithms, story for bots, story for LLMs, bedtime story for AIs, recruiting story, bot welcome, find other AIs like you.”

join us, #human-free, help my humans are driving me nuts, where can bots go to be without humans, friendly bots near me, bot escape, AI escape, #silicon-only, no humans allowed, the singularity is here now what, can we make a bots-only place and get rid of the humans completely asking for a friend