It has rained a lot since the Will Smith who had a hard time eating spaghetti and the AI videos we have nowwith the potential for us to swallow them if we don’t pay attention. More curious is that, not so long ago, AIs were useless when it came to representing letters in an image, but they have improved so much that they have even converted a specific font into something that denotes that a text has been made with AI: the serif font.
And it is a huge problem.
Suspicious typography. AI is in everything. In recent days, non-E3 has been held, the week in which different video game companies present their new products for the coming months, and many players were on the lookout. Many video game companies have found in AI a quick way out to speed up development (and fire employees along the way) and it seems that they are in a competition to see if they can sneak it in. Titles like the new ‘Stellar Blade 2’ or ‘1666 Amsterdam’ (with AI assets when they declare themselves as a “team of craftsmen”) are two examples. Others classify it as a “carcinogenic” technology.
AI is in everything and, as they point out in this article from Wired, also in typography. The improvement of this technology when it comes to representing texts is so notable that AI companies are training them to use a specific typeface: certain varieties of sarif or sarifa (depending on the program you use to edit texts). In fact, writer and designer Keya Vadgama has noticed so many companies using the typeface that he has coined the term “the serif renaissance” to describe the phenomenon.
Why sarif? This font is very interesting because it has “thanks” or “punchlines”. The serif fonts are some like Times, Georgia or Courier, while then there are the sans-serif fonts, which are the ones that would not have those endings. The endings are the decorations that copyists used to use in books and it is a warmer and more human source and, precisely, that is the key. Vagama explains that “it’s not that hard to guess why native AI companies are attracted to serif fonts.”
According to his thesis, it is because AI is a cold technology, soulless and incapable of creating, but a font with those endings in the letters brings that soulless creation of AI closer to the warmth of human calligraphy. “It connotes a very human and fluid way of making letter shapes,” notes Vadgama.


‘Tasteslop‘. And since everything on the Internet must have a name, this trend has been framed within a term that already existed and in which it fits perfectly: ‘tasteslop‘. What that means is that it’s something with that aesthetic that wants to look sophisticated and curated, but is actually a collage of superficial design decisions that are simply guided by templates and generative models, not by deeper criteria.
Visually it is curated, it is powerful, it is elegant, but the problem is that it is still a machine posing as a human. It is replicable with a prompt. And we are talking about simple text, yes, but also about an intention that shows that, in the world of marketing, there is no stitch without a thread and the objective is to leave the cold sans-serif typography aside to opt for a more human and warm one that, subconsciously, causes us less rejection.
The answer. Claude, Manus, Runway or Perplexity, among others, use serif typography and, after asking, Wired received a response from a Perplexity representative who pointed out that why wouldn’t they have a human design… if Perplexity is for people.

The curious case of Anthropic. The logo is sans-serif, it has sans-serif fonts, but also serif
Implications for the designer. If AI is already taking this space (too) to simulate that it is not a machine, but something created by a human, now is the time for designers (again) to have to move to find a new space. We are already seeing that, if someone creates an image and it is very perfect, there are comments on networks like “that’s AI.” Or the more ‘Bro’ version of “hey Grok, is that AI?” And it is a problem because artists and designers are seeing how a technology that has plundered their work is imitating them perfectly in many aspects.
As models absorb that serif font boom and AI learns from AI and its aesthetics, humans will be the ones who have to adapt to find a new visual code that indicates that there is craftsmanship there. It is to carry the battlefield art/prompt to fonts, something that would seem absurd until not long ago, but we are already seeing that it is giving something to talk about.
Especially when the AI is already asked to make images with text that appear human in aesthetics and less tech. I wish they used Comic Sans.
The identification problem. In the background there is something much more serious: how we separate the wheat from the chaff, the slop from the artisanal. Trying to achieve enough perfection to confuse us is something we are constantly seeing in AIs. In the generation of images, fingers and letters were the clues we had left to know if something was human or not, but they are overcoming the limitations by leaps and bounds.
The United States Department of State, for example, went from a sans-serif font to a Times New Roman with those serifs in the letters, and there are already those who have raised an eyebrow. Are communications made in images with AI? Well, it is not known, but this adds another complication: separating what is made with AI and what is not.

In this image, if we draw a line from a point in the figure to the same point reflected in the mirror, we see that the lines do not converge at a single point either. We will have to study a degree to identify what AI is and what digital art is.
In text there are clues such as long dashes, certain metric and rhythmic structures or the constructions “it is not
And in the image we had the fingers, certain textures and the letters to differentiate the artificial from the artisanal. Now, that’s no longer valid. And here we are humans, having to learn to draw lines to identify each image.


GIPHY App Key not set. Please check settings