Since my last post about the increasing amount of AI generated infographics appearing on my social media feed, my LinkedIn experience is now inundated with similar graphics. The topics are always print-based, and – had the graphics had been created and curated properly – could be a useful source of information. Unfortunately, most suffer from:

• Graphics that make no sense;
• Graphics that don’t relate to the topic at hand or what was described in an adjacent paragraph;
• Typographical errors;
• Hallucinated words;
• Similar design structure

They are usually accompanied by a text post that contains:

• multiple bullet points with emojis that no human would ever choose;
• “It’s not this… it’s this” style statements;
• Examples always given in threes; and
• Plenty of em-dashes.

While these points above aren’t a guarantee of AI generated content, they are a good indicator of it. It is also worth saying that some AI generated content does contain useful information, such as the original infographic about varying types of black that I’d originally reposted.

Where this frustrates me is these AI tools are being used by people who are seeking clout, but have a limited knowledge of a topic. With the assistance of AI to generate the content and infographic, these creators then post this content as if it is a valuable resource that was curated just for their market and benefit, and that they are now a source of reliable information that should be trusted.

Since my previous article was published, I’d actively been pushing back on the authors of such content. If I saw an infographic with questionable content or issues, I’d respond to the author, but instead of directly calling it out as AI, I’d simply point out the issues in the graphic.

Initially, posters would openly admit that the graphic was AI generated. But as I challenged more posters about their graphics, they either provide no feedback (possibly ignoring it) or provide an answer that has nothing to do with my questions.

If it can be made, we can spot it…

One particular post featured what I considered to be an AI image that wasn’t an infographic, but a quite convincing factory floor. I’ve chosen not to repost the image here as I don’t want readers brigading the original poster.

Several things about the image stood out as either strange, unusual, weird or impossible. For me, some of the tells were:

• Everything about the press was weird;
• A sheet being press-checked had a motion blur on it (though it wasn’t moving), but the loupe is in focus;
• There is a color bar on the long edge, but looking at the press the sheets are feeding through on the short edge;

There are more obvious tells but I’ll cap it at that. There are other tells in this image but these are more subjective and can be explained away as a “staged photograph” or “artistic licence”, such as:

• How well-lit the room is, given the warehouse environment. Not impossible to adjust this in photoshop after a picture is taken.
• The printed sheet being checked with a loupe. There is too much distance between the loupe and the observer that they wouldn’t be able to see anything with the type of loupe being held. Again this can be explained away by the photographer asking for a particular pose.

…but just as AI can create images, AI can spot them too!

So those were my observations. But why go to these forensic lengths when I could either copy the image (or screen capture it) and take it into any of the dozens of websites that claim to detect AI images? They are only a Google search away!

I then decided to do just that, and found that four out of the six sites that I tried said it was likely that the image was generated using AI. Obviously this isn’t conclusive evidence or a “smoking gun” that it IS an AI image.

To be clear, I agreed with the text of the post and said so in my comment – it made some great points. In my mind, the article was let down by the use of the image, because my mind was left with the thought “if the image is possibly AI generated, is the article AI generated too?”.

I keep getting more of it

Unfortunately, this has led to the platform foisting more of this content on me. Its algorithm has observed me reacting to these posts, and because this engagement is keeping me on the platform, I get more of it sent my way from people on the platform that I don’t usually interact with.

A new weapon in the battle

One thing that I had noticed recently while receiving these infographics was that – from time to time – a graphic might have a tiny circle with the letters CR inside. Curious, I clicked on this circle and was presented with a new prompt:

Turns out this is a new feature from the Content Authenticity Initiative. It is an emerging effort but hopefully this does something to stem the flow of a consistent stream of newly generated AI content. If you click on the CR icon, more information is presented about the image’s provenance.

This is not how I want to spend my free time

The reason that I write this blog is to give back to the industry that has become my career. The information that I post is usually inspired by actual events that I was involved with, or tips and techniques that I think colleagues and peers will find genuinely useful.

This extends to my presence on social media platforms, whether personally or wearing the Colecandoo or Adobe Certified Expert hat. I’ve no intention of taking on the role as the LinkedIn “content cop”, nor do I want to be considered as a digital vigilante taking on these particular content creators.

What I want to achieve is a greater awareness in the LinkedIn community that AI driven content is no longer limited to platforms like YouTube or Facebook, it’s now a phenomenon on LinkedIn that users need to pay more attention to.

Soft-Block them instead

Ultimately, my strategy of “call them out and block them” is flawed as the algorithm is just sending more of that content my way. Instead, I’ll pivot to “soft-block them”.

This will hopefully have the added benefit that it changes my algorithm to understand that I no longer want to see this type of content, or content from these particular posters, so that my LinkedIn feed returns to more of the genuine content from real-world creators whose opinions and thoughts are their own.

What are your thoughts?

Am I overreacting to these AI generated infographics and images? What has your experience been like? Is the underlying message of the content more important than how it was presented? What is your opinion of AI within this space? Please let me know your thoughts in the comments section below!