Can AI Have Taste? The Question That Defined Replit's Vibecon in New York City
In the heart of downtown New York City, a conference unlike most tech gatherings took shape. Artists sat next to software engineers. Filmmakers swapped ideas with product builders. And hovering over every conversation was a single, deceptively simple question: can artificial intelligence have taste?
That was the charged atmosphere at Vibecon, a two-day conference organized by Replit — the fast-growing vibe-coding platform that has become one of the most talked-about names in AI-assisted software development. The event billed itself as a gathering for "creatives and AI," and it largely delivered on that promise, blending the worlds of technology and artistic expression in ways that sparked both genuine inspiration and a fair amount of skepticism.
What Is Vibe Coding — and Why Does It Matter?
To understand why Vibecon generated so much buzz, it helps to understand the concept of vibe coding itself. Vibe coding refers to a style of software development where users describe what they want to build in plain, conversational language, and AI tools generate the underlying code. Rather than requiring deep technical expertise, vibe coding lowers the barrier to entry for building apps, websites, and digital tools dramatically.
Replit has been at the forefront of this movement, positioning its platform as the go-to environment for people who want to build products with AI assistance — regardless of whether they have a computer science background. In fact, Replit CEO Amjad Masad has gone on record suggesting it may even be counterproductive to pursue a traditional computer science degree in the current AI landscape, a provocative stance that has earned the company both admirers and critics.
Vibecon was, in many ways, a celebration and exploration of this philosophy. If anyone can build software now, who gets to decide what good software looks like? And does AI have any role in answering that question?
Taste: The New Core Skill in the Age of AI
Inside the venue, the word "taste" came up again and again. It has become a genuine buzzword in AI circles over the past year, with technologists, designers, and entrepreneurs arguing that as AI handles more of the execution work — writing code, generating images, drafting copy — the ability to judge quality, make meaningful aesthetic choices, and know what "good" looks like becomes the most valuable human skill remaining.
The debate is far from settled. Some attendees at Vibecon were optimistic, arguing that AI tools actually amplify human taste by removing technical friction. When you no longer have to spend hours wrestling with syntax or debugging errors, the thinking goes, you can focus more energy on vision and judgment. In this framing, AI is a collaborator that serves your aesthetic instincts rather than replacing them.
Others were less convinced, raising pointed questions about whether AI outputs are subtly homogenizing creative work. When millions of people use the same underlying models, do the results start to look and feel similar? Can an AI, which learns from patterns in existing human-made work, ever generate something that reflects genuine originality — or is it always, at some level, remixing what already exists?
Spike Jonze, Amjad Masad, and the Creative Process
One of the most anticipated moments of the conference was a conversation between Replit CEO Amjad Masad and acclaimed filmmaker Spike Jonze, known for directing visually inventive films and music videos that blend surrealism with deep emotional resonance. Masad, who wore a New York Knicks hat for the occasion, sat down with Jonze to explore how the creative process works and where AI fits into it.
The pairing was deliberately unconventional. Jonze is not a software developer or an AI researcher — he is an artist. His presence at a tech conference underscored Replit's attempt to position vibe coding not as a purely technical discipline but as a creative one. The message being sent was clear: building software can be an expressive act, and the tools that support it should be evaluated by artistic standards as much as technical ones.
The conversation touched on how creators make decisions, how they know when something is working, and what role intuition plays when AI is generating options faster than any human could evaluate them. These are not abstract philosophical questions — they have real implications for anyone using AI tools to build products today.
AI and Creativity: Optimism, Skepticism, and Everything In Between
Attendees at Vibecon left with mixed but thoughtful impressions. There was genuine excitement in the room about what AI makes possible. The prospect of someone with a bold idea but no coding background being able to build and ship a real product is compelling, and Replit has done more than almost any company to make that a practical reality rather than a distant promise.
At the same time, some observers expressed fatigue with the "taste" discourse, suggesting it risks becoming a convenient way to sidestep harder questions about authorship, originality, and what it actually means to create something meaningful. Saying that humans provide the taste while AI provides the execution is a tidy narrative — but creative work has rarely been that clean a division.
What Vibecon Reveals About the Future of AI-Powered Building
Whether or not AI can have taste in any meaningful sense, Vibecon made one thing clear: the conversation about creativity, technology, and what it means to build something in the age of AI is only getting louder. Replit is betting that bringing engineers and artists into the same room is one way to move that conversation forward productively.
With Replit reportedly projecting significant revenue growth and vibe coding continuing to attract mainstream attention, the stakes of these debates are rising. The tools are evolving rapidly. The communities using them are expanding. And the questions being asked at conferences like Vibecon — about taste, judgment, and what humans bring to the creative process that AI cannot — are likely to define how this technology is understood and shaped for years to come.
For now, the answer to whether AI can truly have taste remains open. But asking the question seriously, in a room full of people who care deeply about both technology and creativity, might be exactly the right place to start.