State of the art in AI image generation as we go into 2026.

2»

Comments

  • Igor Petrovic
    Igor Petrovic Posts: 353
    What is interesting for me while interacting with ChatGPT on various topics is its pharisee position on general ethics, when advising one how to deal with real-life problems. While being pretty much workaroundish about his moves :) 

    I do not complain about it, just notice how it kind of resembles a widespread parenting model when some of us were kids 30-50 years ago (do not do what I do, but do as I tell you).
  • Mark Simonson
    Mark Simonson Posts: 1,784
    edited April 23
    Instead of artificial intelligence we get artificial smart ass.  :D
  • Craig Eliason
    Craig Eliason Posts: 1,501
    Chris Lozos said:
    To me, AI is only of value if it can do the drudgery part of the job, not the creative part.
    It feels to me like it may be precisely the opposite: AI spews plausible-but-problem-filled output, and it’s up to the humans to clean it up.
  • Ermin Međedović
    Ermin Međedović Posts: 97

    LLMs produce output that appears plausible but contains underlying issues. Yes. Identifying and correcting those issues is left to humans. Yes. 

    That, however, assumes the user can recognize what’s wrong in the first place. And they can't. For most users, “good enough” is simply enough.

  • Thomas Phinney
    Thomas Phinney Posts: 3,132
    I think a lot of the best of what we are seeing is not “AI generates images for you out of nothing,” but rather “AI riffs on something you give it, in major productive ways.” Not quite as dramatic, but maybe more useful.

    Also, actual software coding is where more of the action has been taking place lately—the big AI companies putting more effort into it.

    I still expect to see AI-generated fonts (or even extensions to fonts) Real Soon Now. This year for sure.
  • Ray Larabie
    Ray Larabie Posts: 1,487
    edited April 28
    That presentation is what brought me back to this thread. Don’t just flip through it; zoom in and read the prompts.

    Here’s a near-future scenario: a new business owner prompts an AI to create a logo and a style guide for their business. Assume the AI is fully informed about all aspects of that business. In the style guide, there are custom typeface specimens.

    The business owner uploads the style guide to an image or video generator and can produce all the material they need without ever touching a font; print-ready business cards, vectors for vinyl-cut signs, video presentations, and so on.

    A typography expert could zoom in and pick out flaws, but the material is satisfactory to both the client and their customers.

    Why would this business need fonts? To the business owner, what they’ve done is cheap and effective.
  • Dave Crossland
    Dave Crossland Posts: 1,530
    edited April 29
    That future is already here, it just isn't evenly distributed...

    I didn't look with a microscope but there was only a single Latin glyph I noticed that was drawn incorrectly (an uppercase G)

    I asked a few friends and the non Latin is still behind, like where Latin was about a year ago. It's questionable how much source material exists for non Latin vs the overabundance of Latin in the world... But ... Tick tock.
  • John Hudson
    John Hudson Posts: 3,654
    edited April 29
    Input and editing is an interesting case. In theory, AI could be live (re-)drawing text as it is input, making fonts unnecessary even in direct text authorship unnecessary. It seems, however, a really inefficient way to go about things. And, of course, there’s a bunch of companies lining up to charge users a subscription to do this stuff, while most of the existing infrastructure of text authorship, editing, and presentation is free or effectively free.

    Hmm. Maybe I can make and market a font that is specially tuned for AI prompt writing. I have no idea what that would mean, but then I have no idea what most font marketing claims mean.
  • John Hudson
    John Hudson Posts: 3,654
    edited April 29
    I did scroll down through all the images in the presentation, and it would be churlish not to acknowledge that the sheer productiveness and surface sheen is impressive. But they are also 100% pastiche of styles of images previously made by skilled human beings. Ironically, the style used for the explanatory image panels is pretty close to what I anticipated in 2023, in the conclusion of my chapter on text technologies for the Bloomsbury Handbook of Global Typography: a pastiche of modernism, lacking any of the spirit of modernism. The bad thing that the machine is very good at.




  • Igor Petrovic
    Igor Petrovic Posts: 353
    edited April 29
    Watching the video, I had a flashback, without knowing why exactly. Then, looking focused, I realized it was caused by one of the typefaces looking familiar to me. I found the evoking screenshot and inspected it, comparing it to my font Zoran.

    I am not stating anything, just don't know what to think.

    This is an unpleasant fog we are wandering in. I have no hi-res image to properly compare. It is not even a fixed font, in the next take it might look somewhat different. Even with differences, you can't tell if it was used, but subtly modulated. There are no clear copyrights or proof of sources used.

    All of that gives a pretty shady sentiment around AI. I don't care about discussing the future of AI, believers vs skeptics, etc. I need more arguments that it is completely legal. Until that is addressed, it is a gray zone business. 

  • James Puckett
    James Puckett Posts: 2,044
    I want OpenAI to send a team to India to spend a year photographing every painted sign, truck, and license plate. Then train a new model on that and see what kind of wild stuff it comes up with.
  • Nick Shinn
    Nick Shinn Posts: 2,346
    edited April 30
    Several years ago, I designed the Bodoni Egyptian® typeface. Sounds like a prompt, doesn’t it?
    Easy enough for AI to mimic, with or without fonts, but I’d like to believe that there’s something in the way that I designed and drew it that represents an inimitable human quality that people who specify type style can relate to. My taste; not to everyone’s liking, but unique and special enough, as determined in a font, with all the specificities of spacing and kerning, and not the result of a user’s prompts spun out by algorithms.

    When I designed Neology®, which mixes grotesque and geometric characters in a quite readable manner, I discovered, in using the typeface for various projects, such as my blog, that I didn’t like the mixture and really did prefer the specificity of the Neology Deco version. From this I conclude that the specificity that a type designer embeds in a particular font—a particular weight, too, not an elastic variable instance—is what makes a typeface useful to the highest, and most distinctive degree. As a conceptual entity that typographers can work with in creating layouts, useful because it has meaning for them, connecting with so many other signifiers of design culture, in their understanding of what that is. And this is surely what art directors want for the brand they’re promoting.
  • John Hudson
    John Hudson Posts: 3,654
    edited April 30
    In more general terms: the form of a text is a cultural product. The media of text manufacture has changed and will change, and so will the form, but a notable aspect of previous changes in text media is how much has been preserved alongside that which is novel. Fonts as a medium may go away, but the typefaces that they implemented will persist because they exist in culture, not in data bytes or photo strips or Linotype matrices, etc.. And people will also need to find a way to implement new typefaces — or whatever term they choose to refer to specific and repeatable forms of text elements — within a fontless medium. Perhaps as training sets for AI text manufacturing?
  • Igor Petrovic
    Igor Petrovic Posts: 353
    edited April 30
    So far during license checks, no client has disputed the font in question. Could be discussions on whether they respected the license or not. But I still haven't got the reply: "That's not that font".

    My main concern at the moment is that AI could become an excuse for avoiding font licensing at all. Not that the user relies on AI, but uses it in a conversation as a theoretical black box, while still using the actual font or its derivative. Anybody can say, even for the live text (webfont), that it was created by AI. How to prove it?

    Now they can download the font from a pirate site, delete metadata, load it into the AI, and say give me a minor derivative. And then what?

    I understand the logic of relying on honest and refined clients. But the problem I have is that we build a discourse where rudeness, disrespect, and negligence become a solid argument. Again, you try to check the license, the user says "That's not the font", then what?
  • Dave Crossland
    Dave Crossland Posts: 1,530
    Igor: I need more arguments that it is completely legal. Until that is addressed, it is a gray zone business. 

    It isn't clear to me; but personally what I read on social media is that the argument that it is completely legal in the USA rests on the US copyright law principle of Fair Use. 


    John: a pastiche of modernism, lacking any of the spirit of modernism

    I recently read about Mark Fischer, who wrote some tantalizing books about how we are "haunted by futures that failed to happen", and that Western cultural recycling is useful for the oligarchical owners of cultural production conglomerates because actually imaging a future means imagining upending the status quo, so they are strongly incentivised to foster repeats, and avoid fast forwarding 
  • John Hudson
    John Hudson Posts: 3,654
    edited May 2
    Fair use in US copyright law concerns the context of use in which copyrighted material is used, not the technology of reproduction or derivation. So e.g. fair use covers contexts like commentary or criticism, education, and satire. If we consider AI as a technology for making derivative works akin to, say, a video editing suite used by a human to cut together a video derived from copyrighted material, then a fair use claim would a) acknowledge that the resulting work is derivative, and b) the resulting work is being used within the kinds of contexts permitted by fair use. So far as I can see, neither is commonly the case for AI: derivation is obscured rather than acknowledged, and derived works are used in contexts beyond those permitted by fair use.

    What this suggests to me is that training of AI on a creative work, e.g. Igor’s typeface, may be deemed fair use — something the courts would need to decide on, not something that should be presumed —, in the same way that it was deemed fair use for search engines to gather data from copyrighted material.* But what the end user does with the output of AI, the context in which that output is used, can’t be considered automatically fair use, any more than the output of any other technology is. So then we have the practical problem of identifying the source of derivative works from systems that obscure their input rather than acknowledging it. Igor has identified something that looks like it might be derived from one of his typefaces. It may well be, as well as incorporating things that the machine has derived from other typefaces, mixed together in differing amounts into something a bit generic.**
    _____

    * I have read some suggestions of technical measures to disallow AI training, akin to robots.txt that in theory tells web crawlers not to include website content in search engine data. Such measures are regularly ignored on the grounds of fair use.

    ** Like a recipe that turns individually tasty ingredients into a bland gruel.😛
  • Chris Lozos
    Chris Lozos Posts: 1,477
    What good is a new typeface if it is not truly a "new" typeface.

Categories