AI-generated posters don’t have to be horrible
The thread's real split is not whether AI can imitate different styles, but whether style variation addresses the deeper complaints: repetition, perceived low effort, uneditable outputs, print-production failures, and…
The thread's real split is not whether AI can imitate different styles, but whether style variation addresses the deeper complaints: repetition, perceived low effort, uneditable outputs, print-production failures, and the economic displacement of human designers.
The author argues that AI-generated event posters are not inherently bad, but the default ChatGPT/diffusion style becomes irritating through endless repetition. To test this, they prompted ChatGPT for a spring fayre poster, got a generic craft-fayre result, then asked for deliberately different aesthetics.
ChatGPT named the second attempt as Bauhaus/modernist and supplied a menu of styles such as Swiss, risograph, cut paper, brutalist, 90s rave, Memphis, Japanese minimal, wayfinding, and letterpress. The author generated examples in several of those styles, noted that context can leak extra text into later posters, and concluded that AI posters can be distinctive even if still recognizable as AI.
They also mention a further option: using Claude or Gemini to generate HTML, PNG, or PDF posters with editable text and layers, and link to a catalogue of 100 poster styles.
Key takeaways
- The main problem with default AI posters is not that each one is terrible, but that the same style becomes irritating after repeated exposure.
- Prompting for specific design movements or aesthetics produces more varied and distinctive poster outputs than accepting the default craft-fayre template.
- ChatGPT can name styles and suggest a broad menu of design directions, helping the author discover labels for looks they recognized but could not name.
- Context contamination is a practical issue: extra text invented in one generation can persist into later posters unless the user starts fresh or edits carefully.
- The author does not aim to hide AI use; the goal is to avoid the ubiquitous look everyone is tired of.
- A further step beyond image generation is using tools that output HTML, PNG, or PDF with real text and layers so posters remain editable.
What people think
Variations still feel AI-uniform to some critics
Several commenters argue the article's alternative posters remain recognizably AI-generated and share an underlying uniformity or smell. They say the issue is not just typography or imagery but a lack of vision and imagination, and that models reach for banal associations such as sakura for a Japanese-minimal style.
Counterpoint: the alternatives beat the old default
Other commenters say the article's later examples are far better than the generic AI slop and that the realistic pre-AI alternative was often WordArt, Comic Sans, or clip art. They argue small local events rarely hired professional designers, so AI can raise the baseline even if it is not high art.
Effort signaling and community trust
A recurring disagreement concerns what AI use signals. Some commenters say default AI posters communicate low effort and even low effort pretending to be high effort, making them less likely to attend. Others reply that graphic-design skill is not correlated with organizing a good event, and that a polished commercial-looking poster might signal a cash grab rather than local enthusiasm.
Practical production problems: sizing and editability
A print-shop worker reports that AI-generated flyers and business cards often arrive incorrectly sized, causing white borders and customer disputes because the model does not understand inches. Another commenter adds that image outputs lack production layers or editable files for InDesign, Illustrator, or Photoshop, making hand adjustment difficult.
Recognition, labeling, and bias
Some commenters argue people often judge AI-labeled work as worse regardless of quality, citing cases where a real Monet was mistaken for AI or where blind evaluation would be harder. Others insist AI slop is genuinely obvious and irritating, and that the Photoshop analogy fails because AI performs the whole activity rather than acting as a human-controlled tool.
Economic pressure on designers and small clients
Commenters disagree about the labor and budget effects. Some say AI is cheaper and often better than budget freelancers, and that small businesses simply lack money for top designers. Others argue clients should hire students or local artists, that AI-generated visuals signal low quality or scam risk, and that AI may replace some design jobs while leaving high-end and trust-sensitive work less affected.
Estimates describe sampled comments, not the whole community.