In today's world, where artificial intelligence technology is increasingly reshaping the paradigm of knowledge production, a profound issue confronts the library science community: as machines can directly generate insights and decisions from vast amounts of data, what will become of the core value of libraries—long-standing knowledge intermediaries? An emerging perspective suggests that the key question is not whether libraries will be "bypassed," but whether they can actively undergo a profound strategic transformation to achieve a "dual extension" of their service focus: forward, extending to the source of data, becoming architects of high-quality data; backward, extending to the application end of machine-generated cognition, becoming navigators of critical thinking. This concept of "dual extension" provides a penetrating path for the reconstruction of library value in the AI era.
The Impact of "First Principles": From the Twilight of DIKW to the Dawn of "Direct-to-Cognition"#
To understand the urgency of this transformation, one must first examine the theoretical impact behind it. Traditionally, the service logic of library science is deeply rooted in the DIKW (Data-Information-Knowledge-Wisdom) model. This model depicts a linear, hierarchical value chain, where human cognition and judgment are the core engines driving the transformation of data into information, knowledge, and ultimately wisdom, with libraries playing a key role in organization, filtering, and dissemination.
However, the rise of AI technologies such as large language models is challenging the foundations of the DIKW model with their powerful pattern recognition and content generation capabilities. They demonstrate a potential for "Direct-to-Cognition," extracting patterns, insights, and even decision recommendations directly from raw data, thereby bypassing traditional human information processing steps to some extent. This phenomenon can be viewed as a new "first principle": in data-intensive knowledge production, machines are becoming the primary data processing entities.
This principle does not signal the end of human roles but compels us to rethink the starting and ending points of value creation. If the starting point of the value chain is raw data, and machines are its primary processors, then the initial quality, structure, and "machine-friendliness" of the data directly determine the upper limits of subsequent cognitive outputs. This is precisely the logical starting point for the "forward extension" of service focus.
Forward Extension of Service Focus: Becoming Architects of Data Readiness#
The "forward" strategy requires libraries to shift from being managers of knowledge "finished products" to strategic curators of knowledge "raw materials." Its core mission is to ensure that data reaches an "AI-ready" state before entering AI models. This involves deepening roles at three levels:
-
From compliance guidance to data strategy consulting. Library services are no longer limited to assisting researchers in writing compliant data management plans (DMPs) but must become data strategy advisors early in research projects. Librarians need to proactively engage in discussions about how data structure, annotation methods, and collection design can better serve future machine learning and AI analysis, enhancing the potential value of data from the source.
-
From passive archiving to active data curation. The function of institutional repositories (IR) will upgrade from "digital preservation" to "AI-friendly data centers." This not only means storage but also involves proactive curation activities such as data cleaning, standardization, and linking (e.g., connecting knowledge graphs and ontologies) to enrich data context and enhance its machine discoverability and understandability.
-
From serving "humans" to accommodating "machines" in metadata strategy. The design of metadata must transcend traditional human-centric retrieval services, adopting semantic standards more suitable for machine interaction. High-quality, machine-readable metadata is a key infrastructure for activating data value and achieving cross-domain data integration and efficient AI analysis.
By shifting the focus forward, libraries strategically deploy their traditional advantages in information organization, standardization, and long-term preservation to the forefront of knowledge production, thereby establishing their indispensable foundation in the human-machine collaborative knowledge ecosystem.
Backward Extension of Service Focus: Becoming Navigators of Machine-Generated Cognition#
As machines produce "cognition," a new and significant service gap emerges—explanation, verification, trust, and ethical considerations. AI outputs are often probabilistic, lacking deep context, and can even be "black boxes." Users facing these seemingly authoritative "insights" urgently need critical navigation and support. Therefore, the service focus of libraries must extend to the backend of the value chain.
The "backward" strategy requires libraries to shift from being "providers" of information to "navigators" of cognition, with the core mission of helping users navigate, interrogate, and responsibly apply machine-generated cognition. This also involves the evolution of roles at three levels:
-
From information literacy to critical AI literacy education. The focus of library literacy education must expand from assessing the authority of traditional information sources to examining the reliability of AI outputs. Content should cover algorithmic bias identification, understanding model limitations, result verification, and triangulation, cultivating users' abilities to collaborate effectively and cautiously with AI systems.
-
From reference consulting to explanatory and contextual services. In the face of complex insights obtained from AI, librarians need to assume the role of "AI result interpreters." This includes introducing and applying explainable AI (XAI) tools, and more importantly, placing AI outputs within the specific disciplinary knowledge systems and ethical frameworks for in-depth contextual analysis, helping users understand their true significance and potential impact.
-
From collection development to advocacy for responsible innovation. As a symbol of social credibility and a neutral temple of knowledge, libraries are ideal platforms for discussing AI ethics and advocating for responsible innovation. By organizing interdisciplinary academic discussions, formulating ethical guidelines for AI applications, and promoting privacy protection technologies, libraries can safeguard the core values of the academic community and society while advancing technological progress.
Conclusion: The Birth of an Ecosystem Hub#
The "forward and backward extension of service focus" is not two separate actions but a complementary dynamic closed loop. High-quality data curation at the front end is the foundation for the trustworthiness of cognitive results at the back end; conversely, critical reflection on AI results at the back end will guide the optimization of data collection and processing strategies at the front end.
This strategic concept of "dual extension" paints a new development blueprint for libraries. They are no longer passive knowledge repositories but actively embedded hubs in the entire cycle of knowledge production. In this hub, libraries connect researchers, data, machines, ethics, and society, ensuring a solid foundation through "forward extension" and a correct course through "backward extension." This is not only a profound reflection and development of traditional library science theory but also a clear and determined path for libraries to achieve sustained prosperity and value manifestation in the intelligent era.