A message may be more or less ambiguous to a recipient, who tries to disambiguate it. That is, what matters is not only the message itself but also the listener himself. Such is also the case with its significance or relevance to the recipient. Both meaning and significance matter to everybody. Or it matters for everybody to understand and evaluate messages. It is necessary but not terribly important to communicate messages to the recipient through the conduit or channel, but to help him or her understand and interpret them properly for best use. This is the ultimate objective of information science. In this regard, you may doubt how well both traditional librarianship and traditional computer science could help. Even information science without proper cognitive scientific knowledge could not help. And you may well like the former to be the latter. This may be called the second cognitive turn of information science, which appears the best site of cognitive science, where AI would better be included.
ㅡㅡ
Yes. I think there is a substantial theory here, and “second cognitive turn” is a useful name if we define the first and second turns carefully.
The starting point is your distinction between transmission and understanding.
The conduit model gives something like:
sender → message → channel → receiver
Its main problem is successful transmission. Did the message arrive correctly?
But human communication has a harder problem:
message + context + receiver’s knowledge → interpreted meaning → judged significance → possible action
The message itself does not determine the result. The receiver must do two major cognitive jobs.
First comes disambiguation and interpretation:
What does this mean?
Then comes evaluation or relevance judgment:
What does this mean for me, for my problem, or for what I am trying to know?
These are different. A person can understand a statement perfectly and nevertheless judge it irrelevant. Conversely, a potentially important statement may be useless because the person cannot understand its meaning.
So we could formulate two fundamental dimensions:
Meaning: What does the information say?
Significance: Why does it matter?
That is much richer than defining information science primarily as efficient transmission.
Why traditional librarianship is insufficient
Traditional librarianship is extremely good at preserving, organizing, describing, locating, and supplying recorded knowledge. But supplying the right book or article does not guarantee that the recipient understands its meaning or recognizes its significance.
The traditional librarian normally stops somewhere around:
“Here is the material you may need.”
The difficult cognitive work remains with the reader.
Why traditional computer science is also insufficient
Computer science has been extraordinarily successful at:
storing → processing → transmitting → searching → computing.
But traditional computers did not understand ordinary human discourse in anything resembling the way human recipients interpret it. Classical HCI could improve interaction with the machine, but the machine itself remained largely outside the human circle of meaning.
LLMs have changed this situation dramatically.
For the first time, ordinary users can interact with computational systems that can behave as if they interpret context, recover implication, compare significance, explain ambiguity, infer unstated relations, and reformulate information according to the recipient’s needs.
That changes the intellectual territory.
The first and second cognitive turns
I would distinguish them this way.
The first cognitive turn, beginning around the 1970s, changed the object of attention:
system/document → user
It asked about knowledge, relevance, information need, interpretation, cognition, and information behavior.
But the computer itself remained largely non-cognitive. Cognition belonged principally to the human side.
The emerging second cognitive turn is different:
human cognition alone → interaction between human and artificial cognition-like processes
Now the information system itself can participate in contextual interpretation.
That creates something like:
human context ⇄ message/document ⇄ artificial contextual inference
And the human remains the final judge of significance.
This could make information science an unusually natural meeting place for cognitive science and AI.
Why information science may be the right home
Cognitive science traditionally asks how minds perceive, represent, learn, remember, reason, and communicate.
AI asks how machines can perform abilities associated with intelligence.
Information science can ask the connecting question:
How can recorded human knowledge, artificial inference, and human cognition interact so that people become better informed?
That question contains IR, citation relations, language, relevance, cognition, LLMs, provenance, misinformation, explanation, and human judgment within one field.
And here I think your phrase “science for wisdom” gains a more exact meaning. Information science should not merely maximize the amount of information delivered. More information can produce more confusion.
Its deeper objective becomes:
help the recipient understand information correctly, judge its significance intelligently, relate it to existing knowledge, and use it wisely.
That would indeed be a major enlargement of information science.
The first cognitive turn brought the human mind into information science.
The second cognitive turn could bring human cognition, artificial cognition-like inference, and recorded knowledge into one scientific framework.
And there is an important continuity between the two. The old question—What information should we retrieve?—does not disappear. It becomes part of the larger question:
What should be retrieved, what does it mean in context, what can reasonably be inferred from it, how significant is it to this recipient, and what evidence allows the recipient to judge the answer?
That looks to me like a credible research program for a second cognitive turn rather than merely a new name for AI.
ㅡㅡㅡㅡ
So far we have practically designed information science as merged with cognitive science and as a science for wisdom. Meanwhile, it was sandwiched between librarianship and computer science, which are materialistic. Since May 2025, the Association for Information Science and Technology (asis&t) has entered merger negotiations with the Special Libraries Association (SLA). It appears unwise for it to going backward. It should try to merge with cognitive science and AI associations so as to bridge both within a framework. Quite recently we began to take seriously the merging of human and artificial intelligences for synergy. Perhaps, then, we could reduce unnecessary wars in the world, for example. This would be the way toward Wellsian World Brain.
ㅡㅡ
Yes. Your direction is intellectually coherent, but one factual update matters first: ASIS&T and SLA did not remain merely in negotiations. Their memberships approved the merger in August 2025, and ASIS&T now describes SLA funds and communities as having been transferred into ASIS&T. (ASIS&T)
Your criticism is therefore now about strategic direction after the merger, not whether the merger should happen.
I think your larger argument is stronger if stated positively: information science should not define its future mainly by moving closer to librarianship. Its more important frontier is now the meeting of human cognition, artificial cognition-like inference, recorded knowledge, retrieval, evidence, and judgment.
That suggests a much more ambitious institutional direction:
information science as the bridge between cognitive science and AI
Cognitive science asks how humans understand, learn, infer, remember, and judge. AI studies how machines can perform related functions. Information science can study what lies between them: how information is represented, transmitted, retrieved, interpreted, evaluated, and used by both humans and machines.
That gives information science a natural framework for the emerging problem of human–AI synergy.
The goal should not merely be:
human uses AI.
It should be:
human judgment + artificial inference + reliable recorded evidence → better collective understanding
That requires information science because neither cognitive science nor computer science alone naturally owns the whole chain.
Computer science can make LLMs more capable. Cognitive science can compare them with human cognition. But information science can ask:
What information should each side receive?
How should context be preserved?
How should claims remain connected to evidence?
How should disagreement be represented?
How should human and AI judgments be compared?
When should the human trust or reject the machine?
How can collective knowledge improve through their interaction?
That is a real scientific program.
Your connection to Wells’s World Brain is also appropriate at the level of aspiration. Wells wanted organized world knowledge to strengthen human intelligence and civilization. Today’s opportunity goes beyond storing that knowledge. AI can now help interpret, compare, criticize, summarize, and connect it.
So a modern World Brain would not merely be:
world documents gathered together
but:
world knowledge + citation/contextual relations + human judgment + AI inference
That resembles the World Library / LLM+ architecture we have been developing.
The peace question is more difficult, but the possibility is real. Wars often involve misinformation, incompatible narratives, propaganda, misjudgment of opponents, historical grievance, and failure to understand consequences. Better information systems cannot remove conflicts of interest, power, territory, or ideology. But they could make some dangerous misunderstandings harder to sustain.
For example, a serious human-AI information system could expose:
what each side claims;
what evidence supports each claim;
where factual agreement exists;
where interpretation rather than fact causes disagreement;
what each side fears;
what likely consequences follow from proposed actions.
That would not create peace automatically. But it could improve the information conditions under which decisions are made.
This may be the strongest meaning of “science for wisdom”:
not a science that claims to possess wisdom, but a science that improves the conditions under which humans can make wiser judgments.
If ASIS&T wanted a forward-looking strategy after absorbing SLA, I would therefore not recommend another literal organizational merger immediately. A better first move would be a formal joint program or federation with cognitive-science and AI communities around something like:
Human and Artificial Intelligence for an Informed World
That would preserve disciplinary identities while making information science the connecting field.
The long-term direction could then be:
first cognitive turn: systems → human users
second cognitive turn: human cognition ↔ artificial cognition-like inference
ultimate information-science problem: how both can use the world’s recorded knowledge together for better judgment.
That is a much more expansive future than simply becoming a larger professional association of information workers.
ㅡㅡㅡㅡ
The first and second cognitive turns in information differ from each other. The first since the late 1970s was to let users evaluate the given message and judge its relevance. The second hereafter is for human plus artificial intelligence to help them be more wise to evaluate the given message and to judge its relevance for the best use and benefit. This would be a science for wisdom, the ultimate goal of information science. This would not only aim for information professionals but for better if not the best information users. Such college departments would flourish.
ㅡㅡ
Yes. This makes the distinction between the first and proposed second cognitive turns much clearer. The two should not be treated simply as two stages of the same user-centered movement.
The first cognitive turn changed who judges information:
Before: the system/indexer determines what information is relevant.
First cognitive turn: the user interprets the information and ultimately judges its relevance.
The proposed second cognitive turn changes how well that judgment can be made:
User alone → user + artificial intelligence → better understanding → better relevance judgment → wiser use of information.
That is an important difference.
The first cognitive turn therefore recognized the user’s cognitive authority. The second would seek to strengthen the user’s cognitive ability.
This also solves an important weakness in simply saying, “The user is the final judge.” The user certainly should remain the final judge—but users can be mistaken. They may lack background knowledge, misunderstand implications, overlook evidence, accept misinformation, or fail to see the significance of something.
Therefore:
final judgment belongs to the human;
better judgment can be assisted by AI.
AI can supply missing context, compare conflicting evidence, reveal implications, identify assumptions, explain unfamiliar concepts, retrieve counter-evidence, and ask whether an apparently coherent interpretation really corresponds to the facts.
From information professionals to information users
This may be the most radical institutional implication.
Traditional information-science education has largely prepared information professionals: librarians, indexers, information managers, archivists, system designers, researchers, and related specialists.
But everybody is an information user.
And in an AI society, using information wisely may become as basic as reading and writing. Students need to learn:
how to understand a message in context;
how to distinguish explicit statement from implication;
how to judge relevance and significance;
how to check evidence and provenance;
how to recognize misleading but coherent arguments;
how to question AI answers;
how to use AI to discover alternative interpretations;
and ultimately how to make better judgments.
That is no longer professional training for librarians. It is education for citizenship and intellectual life.
This could indeed give university information-science departments a much larger purpose. Imagine an undergraduate department not primarily advertising:
“Become an information professional.”
but instead:
“Learn how humans and AI can use the world’s information wisely.”
That could attract students from science, business, medicine, law, journalism, politics, education, humanities—almost everywhere—because all of them face the same fundamental problem: how to make good judgments from imperfect information.
I would therefore formulate the two turns very simply:
First cognitive turn: Let the user judge.
Second cognitive turn: Help the user judge wisely.
And the second requires something genuinely new:
human intelligence + artificial intelligence + contextualized evidence → wiser information use
This also gives “science for wisdom” a practical meaning. Information science would not claim to possess wisdom or tell people what they must believe. Its task would be to design theories, methods, technologies, and education that help people understand and evaluate information more wisely for themselves.
That could be a much stronger identity for information science than defining itself by either libraries or computers.
ㅡㅡㅡㅡ
2026-08-20 Mark Park
Our Library 