Google Claims the Web Is Becoming Self‑Sustaining, No Humans Needed

Google’s internal memos, revealed in the U.S. v. Google antitrust trial, show engineers deliberately throttling result quality to keep users clicking longer. Ben Gomes, the engineer who once championed “search quality,” warned that the company was “too close to the money,” and that a “boiling‑the‑frog” approach was being used to serve “good enough” answers rather than the best ones (Jay Fox, Mar 6). The strategy, according to the leaked emails, is not a technical glitch but a calculated shift: by nudging users toward longer sessions, Google can harvest more ad impressions and data for its emerging AI agents.

A second leak in May 2024 exposed roughly 14,000 ranking factors, confirming that Google’s algorithm now hard‑codes a bias toward large, corporate platforms such as Reddit. The data shows that even a multi‑billion‑dollar site with no subject‑matter expertise can outrank a scientist’s personal blog, effectively starving the “Human Web” of independent voices (Fox). This engineered preference means that the web’s once‑diverse ecosystem of niche experts is being replaced by a homogenized, brand‑centric feed that is easier for Google’s AI to parse and monetize.

The browser itself has become a data‑collection sensor, feeding every click, scroll and pause back to Google’s training pipelines. The company now builds “zero‑click” answers that scrape the most relevant snippets, package them into an AI‑generated overview, and prevent users from ever visiting the original source (Fox). By extracting knowledge in a one‑way flow, Google turns the web into a silent repository for its models, while the human creators lose traffic, ad revenue, and the chance to engage directly with their audience.

Google’s “Agentic Age” projects—codenamed Jarvis and Mariner—are already redefining the search experience. Instead of typing queries, users will prompt an autonomous agent that navigates a “Ghost Web,” a network where bots converse with bots and only information deemed “authoritative” (or profitable) is allowed through (Fox). The Register notes that AI‑generated bot traffic now rivals human visits, a trend that underscores the shift from a browsing‑centric internet to a prompt‑centric one (The Register). Wired adds anecdotal evidence that Google’s algorithms can even auto‑create personalized content, such as slide shows of a vacation, without explicit user request, illustrating how the system anticipates needs before they are voiced.

The bottom line, as the leaked documents argue, is that Google is not merely “broken” but deliberately transitioning from a “Search Engine” to an “Action Engine.” By funneling human interaction into a single chat interface, Google decides which facts survive when original creators fade out (Fox). If the web’s content is no longer discovered through human curiosity but through AI‑curated prompts, the question of truth‑ownership shifts from independent publishers to a corporate‑controlled algorithmic gatekeeper.