journalists

Abstract: Large language models (LLMs) increasingly mediate human communication, from drafting emails to summarizing scientific reports, yet whether they faithfully preserve a speaker's position remains largely untested. We model AI-mediated communication as a two-step generation-extraction pipeline: one LLM produces an argument from a specified stance, and a second LLM extracts the stance from that argument. We represent the pipeline as a probabilistic state transition over five Likert-type stance categories and define the stance preservation rate (SPR) as the average probability that the extracted stance matches the initial stance. Across 112 debate propositions, none of the nine LLMs tested exceeded an SPR of 0.7 under the default configuration. Three drift patterns accounted for most of the drift: polarization, deviation from neutrality, and flipping. Among the mitigation strategies tested, including in-context learning, multiple extraction with shuffled options, assertion, and reflection, only adding medium reasoning effort to a reflection prompt for GPT-5.4 substantially improved the SPR, to 0.775, yet polarization remained the largest pattern, with 0.119 of the transition mass. An exploratory comparison with human extraction on a single proposition suggests that drift arises at both the generation and the extraction stage. These results point to a fidelity gap in AI-mediated communication, with implications for journalism, policy deliberation, scientific communication, and other domains where opinion-laden messages pass through language models.

Abstract: Communication network research explains access to information, patterns of participation, and the organization of public meaning through different relationships and observations. This integrative review connects research topics, knowledge organization, and research practices, using bibliographic analysis of a Web of Science candidate pool to guide selective reading. Three judgments emerge from the comparisons. First, the relevance of a connection depends on the task and the criterion of value: team members anticipate consulting colleagues whose expertise they recognize, while journalists distinguish monitoring sources, using their information, and citing them. Second, commonality at one level can coexist with differentiation at another: audiences share outlets while selecting different articles, and shared issue agendas accommodate different evaluations. Third, some differences remain unresolved: contrasting media-use influence findings cannot be explained simply by whether models include selection and content co-nomination. Reading the uses of homophily, transactive memory, sourcing, agenda-setting, and framing resources clarifies which expectations and observations support these judgments. Selected citation contexts also show how conceptual and measurement resources enter the same argument. The resulting agenda calls for comparisons of relationship types across group stages, source use across reporting tasks, and encountered content with recipients' interpretations. These purposive comparisons establish specific connections among literatures without estimating their prevalence or demonstrating a common causal mechanism.