Why it's now so hip to go analog and how AI is the antithesis of that (www.techradar.com)

Retro tech — cassette tapes, disposable cameras, rotary phones — is staging a cultural comeback as a deliberate counterpoint to the ubiquity of AI-generated content. The article argues this revival isn’t merely aesthetic nostalgia but a rejection of the “competently optimized” outputs of generative models, which tend toward safe, homogenous prose and poreless, mannequin-like visuals. People are craving the surprises, mistakes and tactile imperfections that human-made art carries: the hiss of vinyl, a misspelled name, a slightly off-key note — the very “grime” that gives work personality. AI’s errors are different, often bizarre hallucinations (invented alphabets, mutated fingers, fabricated backstories) that feel uncanny rather than charming. For the AI/ML community this is a signal about user values and model design: polish and fluency aren’t the only success metrics. Teams should consider techniques to preserve or simulate human-like texture — controlled stochasticity, editable “imperfection” layers, or interfaces that make creative iteration visible — while reducing harmful hallucinations. The tension will shape research priorities: better calibration and explainability to avoid uncanny failures, plus tools that let creators reintroduce fingerprints intentionally. In short, future generative systems may need to learn how to be “less perfect” on purpose to stay culturally relevant.