Mark Alfano

Ignacio Ojea Quintana, Marc Cheong, Mark Alfano et al.

Attitudes about vaccination have become more polarized; it is common to see vaccine disinformation and fringe conspiracy theories online. An observational study of Twitter vaccine discourse is found in Ojea Quintana et al. (2021): the authors analyzed approximately six months' of Twitter discourse -- 1.3 million original tweets and 18 million retweets between December 2019 and June 2020, ranging from before to after the establishment of Covid-19 as a pandemic. This work expands upon Ojea Quintana et al. (2021) with two main contributions from data science. First, based on the authors' initial network clustering and qualitative analysis techniques, we are able to clearly demarcate and visualize the language patterns used in discourse by Antivaxxers (anti-vaccination campaigners and vaccine deniers) versus other clusters (collectively, Others). Second, using the characteristics of Antivaxxers' tweets, we develop text classifiers to determine the likelihood a given user is employing anti-vaccination language, ultimately contributing to an early-warning mechanism to improve the health of our epistemic environment and bolster (and not hinder) public health initiatives.

Masaharu Mizumoto, Dat Tien Nguyen, Justin Sytsma et al.

Multilingual large language models (LLMs) face an often-overlooked challenge stemming from intrinsic semantic differences across languages. Linguistic divergence can sometimes lead to cross-linguistic disagreements--disagreements purely due to semantic differences about a relevant concept. This paper identifies such disagreements as conflicts between two fundamental alignment norms in multilingual LLMs: cross-linguistic consistency (CL-consistency), which seeks universal concepts across languages, and consistency with folk judgments (Folk-consistency), which respects language-specific semantic norms. Through examining responses of conversational multilingual AIs in English and Japanese with the cases used in philosophy (cases of knowledge-how attributions), this study demonstrates that even state-of-the-art LLMs provide divergent and internally inconsistent responses. Such findings reveal a novel qualitative limitation in crosslingual knowledge transfer, or conceptual crosslingual knowledge barriers, challenging the assumption that universal representations and cross-linguistic transfer capabilities are inherently desirable. Moreover, they reveal conflicts of alignment policies of their developers, highlighting critical normative questions for LLM researchers and developers. The implications extend beyond technical alignment challenges, raising normative, moral-political, and metaphysical questions about the ideals underlying AI development--questions that are shared with philosophers and cognitive scientists but for which no one yet has definitive answers, inviting a multidisciplinary approach to balance the practical benefits of cross-linguistic consistency and respect for linguistic diversity.