Hilarius Bookbinder wants you to stop saying "AI is just a tool." In a new essay on Substack, Bookbinder argues that this phrase gets thrown around constantly by university administrators and tech companies without anyone actually examining what it means. Case in point: a breathless email from his Provost celebrating Google's AI Professional Certificate as a path to higher productivity. The assumption is that AI extends human capability, like a rake makes yard work easier. But Bookbinder thinks the relationship might be running in reverse.
The essay draws on Martin Heidegger's distinction between two ways we relate to tools. "Readiness-to-hand" is when you use something so naturally it disappears from conscious awareness, like glasses or a keyboard. "Presence-at-hand" is when you step back and examine the tool as an object. Most of our AI discourse, Bookbinder argues, never gets past the first mode. We absorb these systems into our workflows without asking who's actually in charge. He proposes that a real tool should increase personal agency, not diminish it. Rube Goldberg machines, he notes, look like tools but actually decrease agency through absurd inefficiency.
Then Bookbinder gets weird in the best way. He brings in an evolutionary biology idea sometimes called the "nest's-eye view." Richard Dawkins argued that genes build organisms to copy themselves. A critic named Patrick Bateson mocked this by suggesting that if a gene's-eye view works, then so does a nest's-eye view: nests make birds in order to create more nests. The punchline? Some evolutionary biologists think that's actually correct. Burrows and nests can function as units of natural selection. A bird is just a nest's way of making another nest.
Apply that logic to AI. Humans build AI systems. Those systems shape how humans think, work, and create. Every prompt you write, every correction you make to ChatGPT's output, every preference you express through Reinforcement Learning from Human Feedback trains the next version. Researchers Mary L. Gray and Siddharth Suri documented how millions of gig workers on platforms like Amazon Mechanical Turk and Scale AI perform the hidden labor of data labeling for $21/hr and content moderation that makes AI possible. Their agency gets stripped away, turned into a computational resource. So if nests use birds, what exactly is AI using us for?