The great sameness: a comic on how AI makes us more alike (www.itsnicethat.com)

A new comic by Jordan Bolton, published in It's Nice That's Light and Shade series, visualizes how generative AI can produce a “great sameness” across creative work: not through malice but because shared tools, default settings and AI’s tendency to nudge users toward a plausible “correct” output compress stylistic variety. The piece anchors academic findings — a 2024 experiment with 36 ChatGPT users and a separate short-story writer study — which both found that while individuals generated more ideas or more “creative” artifacts using AI, the group as a whole produced fewer semantically distinct outputs. James Bridle’s analogy to architects using the same design software captures the problem: personalization can paradoxically amplify a common voice when many people rely on the same generative priors. For the AI/ML community this matters technically and practically. It highlights the need to design models, interfaces and evaluation metrics that preserve collective diversity: more steerable systems, interfaces that incentivize exploration (rather than templating), calibration of default sampling and temperature, diversity-aware loss functions and group-level diversity metrics. Bolton’s comic is a prompt to rethink which creative tasks to automate and when to trust human instinct — a reminder that generative models should expand individual expression, not flatten it, if we intentionally build for steerability and diversity at scale.