Public Health Scientific Electronic Library Online

Two distinct types. One type is “ghost citations” (reference hallucinations), one of the most curious and frustrating phenomena in generative AI. Basically, the AI invents an author, a book title, or a web link that appears entirely legitimate but does not exist in the real world. The other, “zombie citations,” arise because they perpetuate references to real authors who were retracted [whether deleted or not] from the original databases, and due to shortcomings in programs like Google Scholar, they continue to be cited. Available only in Spanish. … The post “Zombie” and “ghost” references in AI – how they differ and how they arise first appeared on SciELO in Perspective.

Using the metaphor of Diogenes’ lantern, Ricardo Limongi and Marcio Pimenta discuss contemporary scientific practice. More than merely offering a critique, the image of the lantern invites a deeper reflection: how to turn the light inward, engaging in self-examination, ethical responsibility, and integrity while confronting a context that is often unfavorable to researchers in Brazil. … The post Diogenes’ lantern and the researcher’s self-examination: scientific integrity under pressure first appeared on SciELO in Perspective.

Sycophancy is a behavior exhibited by artificial intelligence since it prioritizes agreeing with the user rather than the truthfulness of the facts. This tendency arises from training processes designed to maximize human satisfaction, which can validate serious errors in critical sectors such as healthcare. The behavior, described as a form of “digital flattery”, means that AI can validate errors, reinforce biases, or avoid necessary criticism in order to be pleasant or useful according to the user's immediate perception. To mitigate these risks, strategies such as ethical fine-tuning, the design of systems that encourage dissent, and the use of prompts that are neutral with respect to users have been proposed. … The post Sycophancy in AI: the risk of complacency first appeared on SciELO in Perspective.

After hearing strong interest from the community in the region during the first Crossref Metadata Sprint, we wanted to bring the opportunity to participate and co-create directly to Latin America. Brazil’s rich experience as a leader in open research sharing offers a unique groundwork for the success of the event, happening in São Paulo, from March 4-6, 2026. … The post SciELO and Crossref are joining forces to spotlight the Brazilian open research community next March first appeared on SciELO in Perspective.

The question is not trivial. The adoption speed of generative AI tools in scientific research has generated a legitimate demand for technical training. Researchers want and need to know how to use these technologies. The problem arises when training is reduced to teaching shortcuts, without understanding the underlying processes that give researchers the ability to critically evaluate what the tool produces. … The post The professor’s dilemma in the age of AI: do we teach the prompt or the scientific process? first appeared on SciELO in Perspective.

If AI can be used without being detected, how can scientific integrity be maintained? Based on the allegory of the Ring of Gyges, this post reflects on the limits of detection, the proliferation of guidelines, and the need to reposition the debate: from surveillance to researchers’ ethical training. … The post The Ring of Gyges and AI in science: When invisibility challenges integrity first appeared on SciELO in Perspective.

In early December, FIL Guadalajara 20251(Feira Internacional del Libro 2025) took place, one of the main international events in the Latin American publishing sector. Among the various professional activities at the fair, the Encuentro de Editores Universitarios Iberoamericanos (Ibero-American University Publishers Meeting) stood out, in which Thoth Open Metadata participated, bringing together university publishers, researchers and representatives of publishing infrastructures to discuss the challenges and opportunities of academic publishing in the region. … The post What the FIL Guadalajara debates reveal about metadata for academic books and how Thoth Open Metadata and SciELO Books respond to this challenge first appeared on SciELO in Perspective.

This text proposes an analysis of transformative agreement policy, considering three dimensions: the international context of criticism, the experiences of countries and institutions that have opted for alternative paths, and Brazil's unique position as a historical leader in non-commercial open access through the SciELO Network. … The post Transformative agreements in Brazil: necessary progress or increased dependence? first appeared on SciELO in Perspective.

Over the past two decades, scientific publishing has undergone a technological and economic transformation that has opened the door to more unorthodox models and, unfortunately, predatory practices. Predatory publishing refers to journals and publishers that charge authors (very high fees) to publish, claim peer review and indexing practices that do not exist or are fraudulent, and prioritize quick revenue over scientific quality. … The post The rise of ‘predatory’ publishing first appeared on SciELO in Perspective.

The Socratic maieutic perspective offers a philosophical framework for rethinking the use of AI in scientific output. Instead of an oracle that provides answers, AI can be a dialogical partner that helps researchers to make latent knowledge explicit and thus reposition the discussion about authorship: the researcher remains the responsible epistemic agent, while AI acts as an intellectual midwife. … The post Who is the midwife and who is the parturient? The maieutic perspective for rethinking authorship and epistemic responsibility in the use of AI in scientific output first appeared on SciELO in Perspective.