Gregory's ServerPlease do not add genAI images to punch up your writing. You might think that it adds a nice little bit of visual pizazz to your content-marketing piece, but what you're actually doing is *making it look like content marketing* rather than a useful resource. To the extent that content marketing is an effective tactic, it is because you build trust with the customer by providing them valuable information. A genAI turd plopped on top of your writing is a signal that it will be worthless slop.
 44 comments
 @Wes_Montage @Em0nM4stodon @jalefkowit @glyph Any publication that puts a banner on articles that reads "We couldn't afford to pay an illustrator and won't use generative images that steal their work," would gain my instant subscription. I understand the appeal, I also wish I were a competent illustrator, I also see some genAI stuff that looks kinda neat, I learned about the dollar-bill rule when I was the layout editor for my high school newspaper, I understand wanting to break up big blocks of text with visual interest for lighter writing. This is why, when I can, I take custom photos or include relevant classical art in my blog. Sometimes I'll even just use a stock image. It feels like genAI is like that but more customized. But it isn't. Unless you are a *real* master with these tools there is an unavoidable sheen that they leave on the generated image. It's the smell, if there is such a thing. This is not just me; if you go anywhere that younger people are congregating online, "boomer art" is the *most* polite thing that they call this stuff. It damages your credibility. If you were lazy enough to fake the image, are you lazy enough to fake the facts? It is *much* worse than just having no image at all. This is hardly an original insight. Lots of other people are posting this exact advice. But I want to emphasize it because I just passed on linking to a page for like the 10th time this week because it included a big genAI hero image which looked like absolute shit. Scanning the article briefly it actually looked pretty good, it did not read like LLM slop, but it is a reputational risk to link to something that will give readers that immediate negative impression, and it's not worth it. As a reference, one of the best and most prolific writers that is (relatively) positive about genAI tools has a website that looks like this: https://simonwillison.net . Please observe the amount of genAI art that he is using for punch-ups or hero images @glyph GenAI images are now at the top of most information security blog posts I read, and they're honestly just annoying. The topic or headline drew me in, I want to get on with it and read your post!  @glyph Saw this example of a neat interstitial image that isn't genAI and doesn't require being a skilled illustrator. Parker Molloy used a screengrab from the video that she was discussing, and edited it for visual effect in a really cool way. It takes artistic skill, but not necessarily skill as an illustrator.   @glyph Thanks, I’m going to use “do you know the youngs call this boomer art” from now on. Should be a good argument because the people using slop in this way think they are at the vanguard of tech innovation. (Of course the likely outcome is they will just hate on me for being negative. 🤷♂️)   @makeworld when you are laying out a newspaper or magazine, if you can place a dollar bill on the page and have it overlap nothing but plain text, that is too long a run of text. You should insert a pull quote, a photo, an infographic or something to vary the flow and add visual interest  @glyph Another way to break up your writing is through the use of headings, subheadings, and bulleted lists. Also, callouts and sidebars. All of this can be done with HTML or CSS.  Sources where you can get interesting and free images: Getty Museum Collection: https://www.getty.edu/art/collection/ Public Domain Review: https://publicdomainreview.org/collections Public Work: https://public.work/ PhyloPic: https://www.phylopic.org/ MNAHA Collections (careful, check copyright, not everything is public domain here!) https://collections.mnaha.lu/index/ QIMBY (quality transport): https://qimby.net/ https://morguefile.com/ - Just make sure you don't accidentally follow the upsell links. https://museo.app/ - Searches the collections of a bunch of museums at once. https://etc.usf.edu/clipart/ - Specializes in old-timey line drawings. Sister site has old maps. https://www.davidrumsey.com/luna/servlet/view/all?sort=pub_list_no_initialsort%2Cpub_date%2Cpub_list_no%2Cseries_no - Lots of very high res old books and book covers.  @glyph how long will it take companies to realize this? all it communicates to us is that whatever they're writing is mostly garbage @pinjontall the first few times I saw it, it was a reliable signal of garbage, and I didn't bother to comment, because sure, great, don't interrupt your enemy when they are making a mistake. but now I'm seeing it on stuff which looks like it is written by actual people with some level of care. I'm not sure if this is an artifact of management pressure or social proof or what, but it's bad  @glyph there are many technical blogs I simply can't read anymore cause they do this shit @glyph And while you're at it, don't put emojis in your email subject lines, because I'm going to filter them as marketing spam.  Yes it is a signal that will indicate worthless slop. They should use MORE genAI! So I can get rid of these people and clean my timeline. Worked like magic in the times of NFTs. I am not sad about anyone I lost contact to in this wave.  @glyph everyone can recognise these immediately now, and it just makes you look lazy and cheap  |
@glyph I'm already noticing myself developing a new mutant strain of what used to be called banner blindness, where if I open a page and see a big AI-generated image at the top my brain instantly stops processing anything else that's there