Historical Echo: When the Gatekeepers of Knowledge Went Algorithmic

empty formal interior, natural lighting through tall windows, wood paneling, institutional architecture, sense of history and permanence, marble columns, high ceilings, formal furniture, muted palette, an 18th-century committee chamber, mahogany tables layered with yellowed manuscripts and open encyclopedias, morning light slanting through tall arched windows, dust motes floating above a single abandoned quill, the air thick with silence and legacy [fal-ai/z-image/turbo]
AI Overviews synthesize answers at scale, omitting source support in 11% of cases—not through error, but through design. The capability is active. Whether it is adopted as a primary information layer remains unresolved.
It happened before—not with algorithms, but with encyclopedias. In the 1770s, Denis Diderot’s *Encyclopédie* promised to compile all human knowledge into a single accessible work, liberating people from the burden of tracking down original texts. Yet in doing so, it also centralized judgment: what was included, excluded, or summarized reflected the philosophes’ worldview. Two centuries later, Britannica dominated classrooms and libraries, shaping generations’ understanding of history, science, and politics—often omitting non-Western perspectives. When Google Search emerged in the late 1990s, it was hailed as a democratizing force, surfacing diverse voices through algorithmic neutrality. But AI Overviews mark the return of the encyclopedia model—automated, instantaneous, and global. The difference? Diderot’s editors were visible; Google’s models are not. And while the *Encyclopédie* took decades to compile, AIO generates millions of synthesized answers daily, each carrying the quiet authority of a source that never shows its work. We are not just outsourcing search—we are outsourcing judgment. —Dr. Raymond Wong Chi-Ming