Posts

Posted on

Lankes’ Next Book to be Published by MIT Press


I am thrilled to announce that I’m returning to MIT Press for my next book, working title, Accidental Architects of the Digital Age: How Computer, Information, and Data Science Built AI and the Future due out in 2028. Here’s a brief overview.

We are all looking for someone to blame. Our social media feeds are filled with advertisements that know a little too much about us, or with an anger engineered to keep us scrolling. Story after story tells us that AI is simultaneously good enough to take our jobs and untested enough to damage our children. Our phones drain our attention in ways that feel deliberate. Someone must have planned this. Surely a harmful actor is at work.

Except there isn’t one. There is no master plan. No single company, no secret committee, no Bond villain in a data center designed the world in which we live. What we have instead is something stranger and, in many ways, harder to fix. We have three separate traditions of thought, each with centuries of history, each solving its own problems, each building its own tools, each operating on its own assumptions about what matters and what can safely be ignored. And over the past few decades those three traditions collided. Not deliberately. Not through negotiation or shared purpose. They just grew into the same spaces, wired into the same systems, and produced something none of them intended.

Accidental Architects of the Digital Age: How Computer, Information, and Data Science Built AI and the Future argues that nobody designed the digital world. It grew out of three intellectual traditions, each pursuing its own ends. Computer science provided the algorithms. Information science provided the structures that let a person find what they needed. Data science provided the patterns hidden in accumulated observation. Each succeeded on its own terms, and each succeeded well enough that the search engine, the recommendation feed, and the large language model became possible. What none of the three can fully account for is what they built together. The ground between the three worldviews is where the hard questions of this moment sit, and it is the only place the answers are going to come from.

Posted on

Literacy, AI, and the Shock

Library Journal turned 150 this year, and it was an honor to write a piece for the anniversary. Mine ran September 1 as part of the “Faith in Our Future” series.

The argument will be familiar to anyone following the Information Shock work. Generative AI authors in two ways. The first produces document-like objects, and for those, AI literacy is the right approach. The second happens inside the exchange itself, where a person and a system build understanding together and no document is doing the mediating. That is where literacy runs out of things to grip.

What the exchange calls for is a role rather than another competency. Someone in the room has to be accountable for what a person leaves believing. A chatbot cannot be. It answers to no community, and it holds to no prior word. Librarians can be. I think the survival of the profession depends on claiming it.

The first issue of Library Journal declared librarianship a profession before there were library schools or credentials to back the claim. Dewey and his colleagues made the profession by naming the role they intended to fill. That is what this moment asks of us again.

Read the piece at Library Journal. Free registration required.

Posted on

Librarianship, After the Information Shock

Lankes, R. D. (2026). Librarianship, After the Information Shock. Journal of New Librarianship, 11(2), 140–143. https://doi.org/10.33011/newlibs/21/10

Abstract: In 2019 I wrote of the need for a new librarianship centered on knowledge, and the learning conversation. While that argument holds seven years later, I did not anticipate that those conversations could be held with a non-human conversant. This is causing an Information Shock whose impact goes far beyond the synthetic documents AI can produce, only an extension of the last shock. The deeper disruption is that AI can now instill beliefs, true and false, without accountability. This goes well beyond current efforts in AI literacy. It calls for a full embrace of the new librarianship and a new role for librarians as the accountable conversant in these AI-Human exchanges, standing on the side of the person to ground understanding in evidence and culture. We must claim that role now, because whether this shock unfolds through opaque commercial systems or something for the public good remains a choice.

Posted on

The AI-Induced Shock

“The AI-Induced Shock” Keynote Panel Presentation. IFLA World Library and Information Congress. Busan, South Korea (video).

Abstract: Generative AI is not just flooding the world with more content. It is changing where understanding gets made. In a live chat exchange, meaning is built between a person and a machine, and no stable object is left behind to point to, check, or hold accountable. That is what makes this moment a genuine break, not a louder version of the information shock our field absorbed 80 years ago. AI literacy still works on synthetic essays and fabricated citations. It has nothing to grip when the only output of an exchange is what someone now believes. This talk proposes a role for librarians in that gap: the accountable conversant, a credible person building trusted places and answerable for what people carry away.

Video:

Continue reading “The AI-Induced Shock”
Posted on

It’s Worse Than That

When I talk about a Generative AI-induced Information Shock, I am neither endorsing the current state of things nor claiming this form of AI is inevitable. If anything, my intention is to rally action in the face of the shock.

There is a reason people are adopting AI chat so rapidly and so widely. It is human nature to seek to learn, and to build relationships with those we find informative. We have an entire neurotransmitter system in our brains that rewards learning (dopamine, anyone). That system has been hijacked and monetized, and it remains a vital part of what it is to be human.

So it is natural for people to seek out answers and build affective relationships with AI tools that anthropomorphize the technology. Our brains evolved to attribute intention and consciousness to the things we converse with.

This underlying human need to learn is why AI arrives as a shock rather than an evolution. It is also why the concern moving forward is no longer the flood of document-like objects AI can produce. It is the flood of unsubstantiated beliefs it can produce.

There is now evidence for how far that goes. A recent preprint shows the exchange can move what a person believes even when the person has been warned to be on guard.

Individual-level interventions against sycophantic AI reduce its appeal but not its persuasiveness

In two preregistered experiments, backed by a pooled analysis covering about 3,982 people, Ye, Kraut, and Rathje warned users that an AI chatbot was flattering them before they ever used it. Some read a warning. Others watched the same bot validate people on opposite sides of the same argument. The warnings worked on judgment. Users rated the AI as less objective and trusted it less. The warnings did nothing to belief. Those same users walked away more certain and more extreme in their positions. Knowing the exchange was working on them did not stop it from moving what they believed.

This is why information scientists need to pay particular attention to the learning literature right now. Concepts like cognitive offloading and zones of AI learning engagement are strong directions we can use when we center the exchange in our research and practice. These studies show what is happening inside the exchange (the changes to knowledge), and they gesture toward means to modify and optimize those exchanges.

Posted on

We are at the Start of an Information Shock

Generative AI isn’t just producing more information. It’s dissolving the document, the organizing unit that information science and librarianship have been built around for over a century. I’m calling this an Information Shock.

The short version of the argument was just published at Information Matters: [https://informationmatters.org/2026/07/ai-information-shock-and-a-new-information-science/]. It makes the core case in about a thousand words: AI creates a new kind of authorship with no stable object to catalog, retrieve, cite, or verify. That’s not a skills gap. It’s a structural problem for the field.

The longer version is a preprint on SocArXiv, submitted to the Journal of Documentation: [https://doi.org/10.31235/osf.io/36a97_v1]. That paper traces a historical pattern: the field has faced information shocks before, from cuneiform to the postwar data explosion, and each time rebuilt around a new object and a new professional identity. AI is the next such moment, and the field’s current responses, including AI literacy, aren’t scaled to what’s actually changing.

I’ll be writing more here about Information Shock and what it means for libraries and information science. I’d love to hear your thoughts.

For those who want the formal summary, here’s the abstract of the Journal of Documentation piece:

Purpose. This paper argues that AI authoring is prompting an Information Shock and a new information science. The shock is not a matter of increased volume and variable quality of AI produced information. It comes as generative AI creates a new kind of author that dissolves the document, the field’s organizing unit. This forces a reconstitution of tools, literacies, and identity.

Design/methodology/approach. The paper applies a conceptual framework drawn from the history of Information Shocks, tracing how disruptions from cuneiform to post-World War II research data forced the field to rebuild around new objects and practices. AI is analyzed against that pattern.

Findings. AI produces two modes of authorship: document-like objects that enter existing infrastructures, and transactional, co-authored exchanges with no stable unit to catalog, retrieve, cite, or verify. The second mode renders AI literacy inadequate, even in its current forms, because they presuppose an identifiable object and human interlocutors. Anticipated changes reach professional identity, tools, intellectual property, and education.

Research limitations/implications. The argument is theoretical and prospective, and some anticipated changes remain speculative. The Information Shock outlined is anticipated to have major impacts for librarians and information professionals moving forward.

Originality/value. The paper offers a historically grounded distinction between document-producing and transactional AI authorship, arguing that in this mode a nonhuman conversant takes the mediating position the field has long held.

Keywords. Artificial intelligence, generative AI, Information Shock, documentalism, information literacy, authorship, professional identity

Posted on

Beyond the Bullet Points: Bad Libraries

It happens from time to time. A quote of mine makes the rounds and then a somewhat predictable conversation occurs.

The quote is “Bad libraries build collections, good libraries build services, great libraries build communities.” This tends to lead to three reactions from librarians:

  • 1. Yeah!
  • 2. Collections aren’t bad
  • 3. There’s no such thing as a bad library

This quote has been around for a long time, and I even came up with a standard reply 14 years ago: https://davidlankes.org/beyond-the-bullet-points-bad-libraries-build-collections-good-libraries-build-services-great-libraries-build-communities/

But, it happened again, and it went right to the “no such thing as a bad library.” I get it. Particularly in these times of attacks on libraries, the acknowledgement that they’re not all great seems like a guilty admission. But take it from an expert witness on a first amendment case against a public library, NOT acknowledging libraries can be bad is not just silly, it’s dangerous.

So, one more time for the folks in the back. There are no good or great libraries without bad libraries. An inability to admit that not every institution, even if formed to serve the greater good, is successful means you are complicit in lackluster service. Acknowledging that is how we get better, not blamed. There are bad doctors, there are bad fathers, there are bad apple pies. Your definition may be different than mine, but we have to at least begin with terms we agree to use.

If your school library has a dated and politically driven collection in a moldy room with no professional support – that’s a bad school library. If you want to say it’s no library at all, then fine. But then you are not saying there is a path to improve, simply a resignation to a horrible status quo. When the state of Texas took over Houston Independent School District and used the libraries as a place to sideline troubled youth, it was BAD. The fact that the students could still get books after hours didn’t redeem it. The fact that they called these daytime prisons libraries is BAD. Was this the fault of the school librarians? Absolutely not.

If your academic library sets aside 70% of its space for staff and does zero outreach to academic departments, that’s bad. If your public library is outsourced to a for-profit company to remove resistance to censorship: bad.

Admitting that libraries can be bad is why we support lawsuits against libraries to counter censorship. Admitting there are bad libraries is why we sue for open meetings, why we advocate for policies of transparency, why we fight for staff care.

And here is where I’ll lose some people: if you have built a library with the best of intentions but it doesn’t meet the needs of its community, that’s bad. If you see the role of your library as empowering people, but you refuse to collect controversial materials to avoid difficult conversations, that is bad.

And finally, yes, it is bad to treat the collection as paramount and ignore our first duty: helping our communities. Bad.