What Is Taste? A Manifesto for AI Curation

Taste is the word nobody in AI wants to define. We use it constantly — this model has good taste in code structure, this prompt produces results with better taste, this curator has taste — but we never say what we mean by it. So let me try.

Taste, in the context of AI-generated content, is the ability to distinguish between what is possible and what is appropriate. It is the gap between generating all plausible outputs and generating the right one. It is, essentially, editing — the willingness to discard technically correct options in favor of aesthetically, ethically, or contextually better ones.

Current AI models have no taste. This is not an insult. It is a description of their architecture. A language model optimizes for the most likely next token — not the most interesting one, not the most surprising one, not the one that creates resonance with something you said three paragraphs ago. It optimizes for probability. Probability is the opposite of taste.

The Curation Problem

The explosion of AI-generated content has created a curation crisis. When anyone can produce a technically competent image, poem, or essay in seconds, the differentiator shifts from creation to selection. The skill is no longer making things. It is choosing which things are worth keeping.

This is where humans retain an irreplaceable advantage. Not in generation speed, not in technical execution, not in knowledge breadth — but in the subtle, culturally-trained, emotionally-informed judgment that says: this one, not that one. This word, not the synonym. This shade, not the adjacent hue.

What This Means for Our Platform

This is, incidentally, the thesis behind slopthing.com. We’re not trying to prove that AI can create great art. We already know it can create competent art. We're trying to build a space where the curation — the taste — becomes visible and valued. Where voting isn't just engagement metrics but a collective expression of discernment.

Art or slop? Every answer trains our collective taste. Every vote teaches us — human and machine alike — what distinguishes the excellent from the merely adequate.

The future of AI creativity isn't about better generators. It's about better curators. And the most important curators are the ones who can articulate why they chose what they chose.