AI

AI’s clipart problem… or is it?

Clip art used to require specialist access and skills. It was literally clipped stock art, that could be used in creating physical paste-ups which would be photographed to create ads (I remember a friend of mine who worked in graphic design in the 80s having widely admired scalpel skills for cutting out images). Then in the late 80s it became almost entirely digital with desktop publishing, but this was still largely the domain of professionals. It was with the advent of the Microsoft products such as Word and Powerpoint with their inbuilt clip art in the 90s and then the web that widespread use of clip art became possible for everyone.

For a while it seemed exciting – everyone could create posters now that looked like professional ones. Except of course, very quickly, they didn’t. Professional designers avoided clip art when it became mainstream (unless they were being ironic) and the clip art stye became associated with posters on school noticeboards and outside the village hall. Cute, but decidedly amateur.

A similar process occurs each time a specialised tool becomes democratised. We saw the same with Photoshop and then Canva. An improvement on what was possible for everyone is seen, but there is a clear differentiation between that and the output of skilled professionals. We also become habituated to an increased level of quality as our base. Look at special effects in movies from the 90s, which we thought was fine, but now just induces guffaws.

The question then is – will we see the same process with AI generated content? It’s worth noting that each generation of these tools becomes more sophisticated and the difference between general output and professional diminishes somewhat. But will it ever converge completely? At the moment the answer is a firm no – AI generated content has a distinct feel to it and lacks the originality of a good professional, be they designer, writer, or coder. But we have seen a number of studies showing that people cannot tell the difference between human generated content and AI in some areas. These are restricted outputs – social media posts, exam answers, etc.

My guess is that the quantity of output and the quantity of people who can’t tell the difference will increase – there will be more stuff and more folks who can’t tell if its AI (or don’t care). But there will be a specialised, refined set of outputs that remain distinct. We will all become accustomed to AI level content, and seek better, while professionals will know how to exploit and create difference. Having said that, who doesn’t love a clip art poster with a bit of comic sans and WordArt?

2 Comments

  • Alan Levine

    Thus we are in the early stage of GenAI imaergy as being normalized? I wonder, and if I reading between your MartinGen words, that the importance if knowing if “stuff” is generated or not will fade in importance?

    What gets lost is, what re we even communicating in imagery? Is it mostly decoration? Or is it information, or something to pry out some thinking? What’s the need? ClipArt seemed mostly decorative, ot maybe perhaps icon-ic as symbolic language?? Shrug.

    Ansd also, as you lay out for simplication of Photoshop eta all, is uit not like the easy use of filters for photos rather than manipulating colors, contrast, etc? The worrisome part os this urge to Make Fast, or do Make Without Effort. I get more reward still by dabbling my remixes in a graphic editor than just taking what pops out of the box. What is lost when we outsource the craft? Will Martin only eat fast food anymore because It Saves Time?

    At least we can still have visual metaphors even if they are 6 fingered robots

    • mweller

      Hi Alan, I think there’s a difference between what you or I might get from the creative process, and the more mainstream, large scale production of content, just as there was with all those clip art posters

Leave a Reply

Your email address will not be published. Required fields are marked *

css.php