Internet knowledge creators are at a crossroads. Independent writers, recipe bloggers, wellness podcasters, and parenting experts who’ve spent years building their own online knowledge bases face what appears as a binary choice: compete for attention against AI-generated content or sell their carefully crafted work as training data to large language models.

On one side, creators watch generative models (often trained on scraped web data) produce endless streams of content at a fraction of the cost. Some of this might be slop, but some of it is quite good and only getting better. Soon AI-generated content will outnumber human voices online. This isn’t necessarily a bad thing and could actually be quite a good thing.

But we must protect human-made, trustworthy forms of knowledge to anchor our online discourse. If creators continue trying to compete with algorithms through purely traditional formats (books, blogs, podcasts), they may exhaust themselves to obsolescence — leaving us with fewer reliable human voices to ground our understanding of truth.

On the other hand, surrendering to AI comes with its own problems. Following established publishers like The Atlantic and Vox Media, creators could choose to monetize their existing content by selling it as training data to AI companies. While this might provide a short-term financial boost, it’d mean surrendering years of work to feed the very systems that could eventually replace them.

A third path is emerging. One that doesn’t force creators to choose between resistance and capitulation. Imagine a scenario where knowledge creators keep ownership of their data while embracing AI by creating their own generative interfaces to connect with audiences.

New platforms are springing up that value expert knowledge — from Gigabrain's search engine for user-generated content to Dewey's AI research librarian. Dewey works on behalf of creators by curating and sharing their knowledge with audiences, much like our USV librarian. I imagine third path platforms resembling a kind of agentic Patreon or a custom ChatGPT for human-made knowledge.

I’m excited about how AI creating deeper, two-way exchanges between creators and the public. Novel experiences like Notebook LM’s generated conversations suggest we’re on the verge of a new communication modality (what’s next after podcasts?), and knowledge creators are well-suited to lead with access to the right tools.

As AI increasingly mediates access to our information, the role of trusted human experts will become more crucial, not less. The future looks like creators combining their expertise with AI in ways that preserve humans’ most valuable assets: their relationships and their ideas.