Lab

AI Exploration
AlienInterface
Einleitung
Semiotik

AI Explorartion

AI can do everything—so why create pictograms yourself? Does AI even understand the concept of pictorial signs? How accurate are the results? Where can AI provide support? A small test.

The following is not about proving that generative AI is incapable of something or makes mistakes, but simply about testing where the possibilities and limitations lie. Of course, I must accept the criticism that some of the prompts are not particularly well-developed—but that may be acceptable. Tools: Midjourney, DALL·E, re:craft.

#01 A Set of icons for Communication

#02 A Set of pictograms for Communication

#03 A Set of icons for Communication, black and White, outline

#04 A Set of pictograms for Communication, black and White, outline

#05 An Icon for “Chat”

#06 An Icon for “Woman”, black and white, outline

#07 A Pictogram for “Woman”, black and white, outline

#08 A simple iconset for a hotel, black and white, outline

#09 Trashcan pictogram, flat, yellow, black, outline

#10 An index for “right”, black and white // Semiotic isn’t a thing … 

#11 Pictograms in an urban environment // Pictograms in a rural environment

#12 Fight, flight, stop, start, light, dark

#?? Just kidding 🙂

Conclusion

AI can do everything? Not quite—and that was hardly to be expected. Rather, one would have to train a dedicated model, since this is a niche topic that, due to countless download options, is either not needed or has so far been considered of little use. It may help as inspiration or in searching for motifs—but even here, in my opinion, its usefulness remains limited.
Although there are now also specialized models, these too tend to produce fairly generic representations that will stylistically fit hardly any project. So—better to do it yourself.

Home

Book

Lab

Theory

Play