The Information Shock from the wide availability of generative AI is not a product of the hype surrounding commercial AI tools. The hype is a product of the shock and decreases the predictable future. The ripple effects of the shock will continue even without some mythical artificial general intelligence or Singularity. It is caused by the wide availability of a non-human, stochastic conversant that ultimately dissolves the concept at the center of information science: the document. A conversant is a participant in an exchange, the other side of the back-and-forth through which understanding gets built. Two people talking are conversants. What is new is that one of these conversants can now be a machine, and the exchange, not any document it produces, is where learning happens. These conversants need not be conscious to participate in the conversation; the capacity for exchange is what qualifies them. The exchange, not the document, is what our tools will need to describe and hold accountable.

If you’re feeling off-balance about the rapid rise of generative AI, that is not a personal failing. Things are uncertain and they will require our field to develop new tools, a new central object, and, in the end, redefine our field. Just as in previous shocks (from scribe to librarian to documentalist to information scientist).

Let me take these points in order. First, an Information Shock is when a new technology or societal movement significantly disrupts the knowledge infrastructure. That infrastructure is composed of people seeking and sharing information; the technology they rely upon to do so (printing, TV, the web); the sources of stored information used (libraries, archives, web sites, data stores); and finally, the policies that make up the rules of who is allowed to share what.

At the end of World War II, the technologies of information shifted from paper to microfilm, and eventually from analog to digital. Masses of research and technical literature flowed into a ramped-up federal research program consisting of scientific documentation recovered from Axis countries as well as the substantial scientific mobilization of the Allies (for example, the Manhattan Project). As the post-war period progressed, the vastly different information-sharing policies of the West and the Soviets became a defining character of the Cold War. Disruptions were felt across the whole of society, all turbo-charged by the exponential growth of computing capacity.

That Information Shock was one of many in history. Here are a few illustrative moments, not by any means a complete list, to show the scale. Ancient Alexandria wrestled with one when documents were first centralized en masse. The printing press reshaped scribal culture into a mass-production one, and with it the economics and politics of who could publish. Another shock occurred with the influx of recovered ancient texts into medieval Europe, which helped push forward the universities and the Renaissance.

That is the frame of the current shock. The mechanism causing the shock is generative AI, and it is doing two very different things, and one is a real problem for the field.

The first is producing document-like objects: books, articles, images, code. This is the visible problem. Slop floods book shelves and feeds, synthetic articles enter scholarly databases, citation practice degrades. But a synthetic book still behaves like a book. A cataloger can still describe it and a reader can still check its claims against the sources it names. Our tools bend around it. And if this mode were all there was to the authorship claim, the strongest objection can be offered quickly: the model is trained on human text, so what looks like authorship is really recombination at scale. That is a genuine problem the field has been handling for decades, however. It is not the shock.

The second mode is where the shock lives. Open a chat window and something different happens. You and the system build a piece of understanding together, in real time, and what you build is not a document; it is knowledge. Knowledge is built in exchanges, not stored in documents. Even Google no longer just hands you results. It talks back.

And as important, the exchange is unrepeatable. Ask again tomorrow and the answer shifts. You can save the transcript, and the transcript is a real artifact, but it is not what mediated your learning. The learning happened inside the back-and-forth. The difference is not just the mode of learning (a conversation), but that one of the conversants is always available with a knowledge base trained from the deconstruction of billions of documents and data points.

None of this requires the AI hype to be true. Grant the skeptic every point about the valuations and the breathless anticipation of the Singularity. The shock still operates on the ground. Pew put numbers on this earlier in 2026: about half of American adults are already using chatbots, and when they run a search, most of them read the AI-generated answer at the top of the page. Students are writing essays with these systems. Doctors are using them to triage patients. Politicians have already put them to work manufacturing propaganda. None of this is waiting on artificial general intelligence to arrive, and stock valuations could collapse tomorrow without changing what is already in wide use on the ground.

The field of information science was formed on measuring how well a system handled a body of documents: recall, precision, and relevance. None of these metrics can take the measure of a conversant. What matters with a conversant is trust. Part of our new task will be to operationalize credibility as the reliability of knowledge changes in an exchange, not as a property of a document.

The new post-shock information science that follows will not look like the old one with AI bolted on. It will start from a different object. Where the document was the stable unit our tools organized around, the exchange is what our tools must now describe and find ways to hold accountable. Where information retrieval was our signature technology, something closer to conversation architecture will take its place. And where information scientist was the identity forged in the last shock, another title is likely to emerge from this one, just as information science itself succeeded documentalism and the American Documentation Institute changed to the American Society for Information Science.

So again, if you’re feeling off-balance about the rapid rise of generative AI, it’s not a personal failing. It is a shock, and shocks are conditions a field works through over time. This one is ours to work through, and to shape.

Author’s notes: The author used Claude (Opus 4.8) for editorial assistance, including review of the article’s logic and identification of errors. The author evaluated, edited, and accepted or rejected all suggestions. The argument and conclusions are the author’s own, and the author takes sole responsibility for the content.

The author also thanks Michael Eisenberg, Emeritus Dean of the University of Washington’s iSchool for his feedback.