Investigating the Impacts of Generative AI on Information Seeking

URL SCAN: Investigating the Impacts of Generative AI on Information Seeking
FIRST LINE: # Computer Science > Human-Computer Interaction

The Dissection

This submission studies the trust theater surrounding AI-mediated information seeking: how fluent language, expert signifiers, and procedural changes cause users to treat a probabilistic system as an authority. Its real object is the social manufacture of credibility. That is legitimate, but it remains interface-level analysis. The abstract never reaches ownership, compute concentration, labor displacement, or the competitive forces that make AI adoption compulsory.

The Core Fallacy

It treats the central danger as a problem of perception, procedure, and vocabulary—something that can be described, analyzed, and mitigated through better cross-disciplinary understanding. Under the Discontinuity Thesis, that is downstream. Once AI delivers cognitive work at lower cost and superior scale, institutions cannot preserve human-only information domains, and those who once searched, synthesized, verified, edited, and distributed knowledge lose productive necessity.

A population can correctly distrust AI and still be forced to use it because AI-capital owners control the cheaper channel. Exposing simulated expertise can reveal the mask; it does not stop the machine wearing it.

Hidden Assumptions

• Shared vocabulary can produce meaningful mitigation rather than merely better descriptions of a process driven by competitive adoption.
• Users and institutions will retain enough discretion to choose human-led information procedures.
• Human values, experience, and expectations will continue to govern the procedure after AI becomes the default interface.
• The principal danger is misplaced trust in AI authority, not the enclosure and automation of search, synthesis, gatekeeping, and verification.
• Communication communities can manage the transition without confronting who owns the models, compute, distribution channels, and resulting authority.
• The decisive issue is not infrastructure control but how people label AI expertise.

The supplied abstract provides no evidence of measuring adoption, displacement, ownership, or institutional counter-power. Its scope is therefore narrower than its title suggests.

Social Function

Partial truth serving transition management and prestige signaling. It identifies a real mechanism: linguistic fluency manufactures authority. But the proposed response—cultivate vocabulary, convene discussion, analyze risks, and mitigate practices—turns a transfer of epistemic and economic power into a manageable discourse problem. Institutions can appear vigilant while the underlying adoption logic proceeds. The critique becomes ideological anesthetic when naming AI’s credibility tricks is mistaken for control over AI deployment.

The Verdict

The paper dissects the mask, not the takeover. It is useful as a forensic account of how AI earns trust, but systemically incomplete: it treats authority as a communicative convention when decisive authority will be enforced by cost, scale, and infrastructure ownership. Under DT logic, it does not challenge obsolescence. It describes the persuasion layer through which obsolescence becomes acceptable.