Almost since the moment ChatGPT hit the million user mark, people have wondered if and when the internet will be overrun with AI-generated content. Eventually that concern coined a phrase for the content: slop.
In June, Cloudflare shared that bot traffic had officially overtaken human traffic for the first time. While some social media companies have traditionally thrown up their hands at the implication of bots generating content, others are fighting back against the slop. Late last month, LinkedIn rolled out a “Seems like AI slop” option for all posts. Substack has added a tool for checking if the content you’re reading is AI generated. YouTube has also said it will automatically detect and flag AI-generated content, rather than relying on creators to self-identify the content.
Despite the efforts, AI is continuing to replace classic creative investments like writing and art in our workflows. Eventually, these whole workflows may find their way to an agentic AI, which further reduces the opportunity for human intervention. It will take a lot of flagged posts to win the war against slop.
Personally, I’m especially offended by AI-generated images that are often poorly done, and frequently take multiple iterations to develop. That also means excessive resource waste in the service of a mediocre, vaguely-relevant graphic. Be cool: stop the slop.
Summer Schedule
Bring your tech troubles to the Coven Cary for Open Office hours. We’ll be available August 20 for help.
Everyone needs a break. If you’re a client looking for support, please remember to email [email protected] for support so if one of us is away, another team member can jump in. Thank you!
Have you scheduled your Annual Review? Marit Digital clients have an annual review of their website(s) and/or productivity platform (Google Workspace, Microsoft 365) every year. If you haven’t scheduled yours, use the link below.
And if you have a site or service that needs a review, get in touch!
Upcoming Events
Open Office Hours, August 20
The Coven Cary
BLD Southeast, September 16-17
Washington D.C.T
What We’re Reading
More light summer reading! Well, it’s reading and it’s summer.
- Worse on Purpose
This isn’t an article, but it has articles. Get case studies of companies that made decisions that made their company worse. The site includes the Ledger, which tracks the declining quality of some well-known brands. - How Claude’s text watermark works from Anthropic
Speaking of fighting AI slop, Anthropic has started “watermarking” its output. Good for finding slop, but this also means that any content coming out of Claude – whether it’s something you wrote and had the AI review, or something the Claude wrote wholesale – will be designated as AI content. - Putting sign language AI into users’ hands from Google DeepMind
Deaf users have been unable to take advantage of audio technology, including verbal commands and recording. Only now has Google successfully created a SL2T (sign language 2 text) solution for these users. An interesting read. - Get both sides: I was interviewed by an AI bot (from The Verge) and AI avatars enter the job interview (from Semafor)
I actually read yet a third article that wondered at the future (which is apparently already here) of AI’s interviewing AI’s, especially as it’s documented that AI typically prefers AI responses. Some people wonder if they have to use AI avatars to get past first-round AI interviews. It’s getting dystopic up in here.
Doors are very powerful things. Things are different on either side of them.
– Howl’s Moving Castle, by Diana Wynne Jones
