AI Finally Gets a Contest Betting on Ideas, Not Big-Tech Backing

This QbitAI article focuses on the "build in bilibili · AI Creation Open Contest" launched by Bilibili, interpreting it as a new type of competition distinct from the traditional AI hackathon. The article notes that the contest begins on June 5, with submissions and coin-tossing tallied through August 20—a period spanning 11 weeks, suitable for an AI product to gradually take shape through idea, prototype, user feedback, and version iteration. Its biggest difference lies not in technical difficulty but in greatly lowering the barrier to entry: there are no restrictions on age, education, occupation, or experience; individuals, small teams, and creators with zero background can all participate. The only core requirement is that the final deliverable must be a runnable, interactive product prototype or software application, rather than simply using AI to generate images, videos, or copy. The article argues that as Vibe Coding and large-model tools have become widespread, the hard technical barrier is being flattened, and what truly matters becomes whether one can discover specific problems, define pain points, and turn ideas into something users can validate. The contest also sets no fixed tracks such as education, healthcare, or office work, leaving room for more lifestyle-oriented and personalized needs to be proposed. Examples cited in the article include a plugin that can transport web text, images, videos, and prompt materials across AI creation tools, and a participant who wants to build a Dota 2 beginner 1v1 AI voice assistant. Another key point is Build in Public: participants must post their creation process as a serialized series on Bilibili, where viewers can raise needs and bugs through danmaku (bullet comments) and comments, so that the product is no longer just a result presented on the final day, but is gradually refined through community interaction. In terms of judging, the article emphasizes that Bilibili makes coin-tossing a key signal—the top 10 by coin count can be shortlisted and receive prize money, effectively handing part of the judgment power over to real user behavior. Overall, the article is not reporting the release of a new model or tool, but commenting on a more democratized mode of AI creation and product validation: ordinary people can use AI to turn everyday ideas into products, then hand them over to the community for cold-start, co-creation, and filtering.