I Ran 255 AI Agents to Make Three Comics. The Line Between Slop and Good Is (thoughts.jock.pl)

In a recent experiment, an AI developer used 255 AI agents to create three distinct versions of a comic book, exploring the concept of "AI slop"—the low-quality output that stems from minimal human involvement. The first comic showcased a perfectly executed pipeline that resulted in a soulless, generic product, revealing that sheer scale without editorial input led to a lack of creativity. In contrast, the second version included a critical feedback loop but still felt lifeless, demonstrating that simply upgrading the medium or adding a critique once wasn’t enough to transform the work. The final comic saw a dramatic improvement after the developer invested more time into crafting a detailed creative brief and implemented multiple judgment layers in the production process. With increased human engagement in defining artistic direction and overseeing revisions, the output shifted from mediocre to a compelling narrative, highlighting that the quality of AI-generated content hinges on the human element in the creative process. This experiment underscores the importance of intentionality in prompt creation and the necessity of editorial oversight to elevate AI-generated works beyond mere "slop."