Ever see a bizarre post with a comments section full of people spamming emotes or otherwise responding in a way that suggests they read a description of the post, but didn’t actually see it? Of course interaction bots have been here for a while, but now with AI art (rather than stolen art) it becomes obvious these are actually bots and not people.
What Is An Interaction Bot?
Firstly, in this area, ‘bot’ refers to a bit of code that does something. What the bot does depends on its creator’s goal – some bots sit and ‘watch’ videos to boost view count, others scrape data from websites to analyze it, and some do things like scroll, interact with buttons, and leave simple, plausibly human-sounding comments on posts online. An interaction bot is meant to be a substitute for real human interaction on a post. Since many social media sites now offer moneymaking opportunities based on views or likes, and since everyone likes feeling popular, this is a problem that said social media sites have been fighting since internet points were invented.
Every time some new ‘tell’ makes the bots easier to purge, the bot makers come up with another way to thwart moderators. When bots were getting too specific with likes, the bot makers told them to like a handful of other posts before they started interacting with the desired post, and to stagger when the interactions happened so they didn’t all hit at once. When the comments got too repetitive, a library of comments scraped from places like Reddit started re-appearing in comment sections. It’s easy to borrow human habits, and we’re at a point where an uninterested user is borderline indistinguishable from a bot pretending to be a human, at least just by looking at their browsing habits.
The goal of some bots is to get a lot of followers to follow one account so that account can then be used to sell the new followers something, whether that be a political belief or an actual product. Even on services where views are not tied to money, those eyes are still useful. The way most algorithms work, a popular post becomes more popular because the website shows those popular posts around to new people who might not have seen it. It does this because the popular post in question created engagement, and if the website can keep you engaged, you’ll stay on longer and see more ads. Having bots enter this ring and artificially boost the popularity of certain posts has resulted in a strange new kind of post dominating Facebook. Where a post had to be written by people, and a picture had to at least be stolen from a real person in the past, the widespread availability of ChatGPT and image generators makes some of these fake posts stick out like a sore thumb.
ChatGPT and Image Generators
You can tell a bot to ask MidJourney or Dall-E to generate an image, and then put that image into a Facebook post with a caption you pre-wrote. Once you set it up, you don’t even have to check on it. Once the post has been put up, other bots show up to comment on it or like it, whether they’re yours or someone else’s.
This has resulted in posts like Spaghetti Jesus or The 130 Year Old’s Peach Cream and Filling Birthday Cake getting hundreds of comments all saying “Amen!” or “Looks Good!” with maybe a dozen people asking what everybody is talking about, because the picture usually looks terrible and fake. This isn’t a case of tech-illiterate folks seeing something obviously bizarre and giving it a ‘like’ anyway – these people don’t exist. The better ones may get a couple of real people, but the strange ones are certainly not (look at these pictures The Verge has collected as an example: https://www.theverge.com/2024/4/15/24131162/ill-see-your-shrimp-jesus-and-raise-you-spaghetti-jesus-on-a-lambo ).
We’ve circled around! This new generation of bots are so advanced that, when given the chance to show off the state-of-the-art tech entering the market, they do it without question and accidentally pull back the curtain in the process.
What To Do?
Unfortunately, managing this issue as a user on the web is basically impossible. Even if you keep bots from following your accounts, you’re not immune to seeing bot-run accounts when you’re searching or scrolling. Instead, the best thing you can do is just refuse to engage with engagement bait – when something asks you to say “Heck yes!” in the comments, or leave a like if you love X hobby, you can ignore it, and avoid accidentally propping up bot accounts trying to get big. As for imagery, the bizarre spaghetti creatures and uncanny peach cake bakers are only going to get better – we’re entering a phase of the internet where pictures must be assumed to be fake and verified before they are treated as real, the opposite of what most internet users are accustomed to. On forums like Reddit or Tumblr, a user must look at the comments before taking a post as fact, because upvotes and comments are not necessarily the sign of quality they used to be when the internet was young and lacked bots. It’s a strange new world out there, and the bots are part of it now, for better or worse.