Perfectly Imperfect: A Newsletter About Celebrities' Unusual Tastes Against 'AI-Slop'

Perfectly Imperfect newsletter has become an antidote to AI-slop: in July 2026, British Guardian columnist Emma Beddington described how a daily newsletter (pi.fyi) saves readers from content that neural networks and algorithms polish into faceless mediocrity.

What is Perfectly Imperfect?

The newsletter's concept is deliberately simple: one public figure per day — and a list of what they like right now. No editorial selection in favor of "useful" or "viral." No sponsored inserts, no narrative of "best for your life."

The newsletter's audience consists mainly of American artists and musicians, not necessarily famous outside their niche. But sometimes issues feature names of global scale:

• Francis Ford Coppola — Hawaiian shirts and halva
• Kylie Minogue — Japanese washi paper tape and fresh wasabi
• Lena Dunham — pig training (exactly that, no metaphors)

Other documented tastes include: a cocktail made from Aperol, milk, coffee cream, and olives; the habit of cracking joints against the jaw; a book on sports psychology; the practice of calling strangers without a pre-formulated purpose.

Beddington — a fifty-year-old journalist with a British frame of reference — honestly admits: most names in the newsletter are unfamiliar to her. But that doesn't prevent reading: the preferences are so specific and strange that they create a sense of a real person, not a content-machine character.

Why does the author contrast this with AI content?

Beddington calls the modern internet a "depressing cesspool" — a place where every website visit results in "a cortisol jab straight to the eyes" or AI garbage interspersed with protein ads.

"The more I learn about celebrities' strange hobbies, the more inspired I become.

So much for AI supposedly flooding us with faceless taste slop," the author writes.

The term "tasteslop" from the column's subtitle is the author's derivative of "slop" (AI garbage) applied to taste perception: algorithmic recommendations are optimized toward statistical mediocrity, and all content starts to "sound the same." Perfectly Imperfect works the opposite way — the value of content is directly proportional to its strangeness and impossibility of scaling.

An important caveat: the column is not a technological critique of AI. It's a journalist's personal essay about what she reads with pleasure. Nevertheless, it formulates an observation: reader interest in obviously human, imperfect content is growing precisely against the backdrop of increasing machine smoothness.

What this means

Beddington's essay registers a small but indicative cultural countermovement. As the volume of AI-generated content grows, part of the audience begins to perceive "too polished" as a sign of inauthenticity — a signal that there is no living person behind the text or recommendation. Formats with obvious human roughness — niche newsletters, personal quirks, uncomfortable preferences, admissions of absurd hobbies — become a steady bet against the algorithm.

For teams developing AI products and building content strategies, this is a signal: automated perfection is not always perceived as a competitive advantage. Sometimes the reader wants exactly that crooked line that a machine would never draw.

What is AI-slop and why is it a problem?

It is content polished by algorithms into faceless mediocrity. The Perfectly Imperfect newsletter counters this by showing the real tastes of public figures without editorial selection and sponsored inserts.

What is AI-slop in content?

It is content that neural networks and algorithms polish into faceless mediocrity — uniform, predictable output of automated processing, devoid of originality and human taste.

What is the Perfectly Imperfect newsletter?

It is a daily newsletter where one public figure per day shares a list of what they like right now. It contains no editorial selection in favor of "useful" or "viral," no sponsored inserts, and no narratives.