AI in the Research Lifecycle

Abstract

The Workshop on AI in the Research Lifecycle was held on October 21, 2025, at the 13th Technological Ecosystems for Enhancing Multiculturality (TEEM) conference, which took place at the Research Institute for Educational Sciences (IUCE) of the Universidad de Salamanca from October 21 to 24, 2025, and lasted one hour.

AI in the Research Lifecycle offers a critical, practice-oriented tour of how generative AI (GenAI) can responsibly augment research from ideation to dissemination. The session begins by situating GenAI as the branch of AI driving today’s disruption and clarifies why its everyday integration represents a qualitative inflection point for knowledge work. It then sets a clear objective: to foster ethical, well-informed, and productive research practices when deploying GenAI.

A unifying framework maps GenAI support to each stage of the research cycle (idea formation, proposal writing, state-of-the-art reviews, data collection, coding and analysis, reporting, publishing, and communication) while emphasizing that prompt quality and contextual grounding are decisive for output quality (illustrated by the prompt–context–response schema). Concrete exemplars show how to: 1) brainstorm and structure objectives and hypotheses; 2) interrogate papers with targeted questions; 3) run “deep research” workflows for evidence-bound drafts; 4) convert and manage references (APA/BibTeX); 5) analyze public datasets with transparent code; and 6) generate outreach artifacts such as spotlights, slides, and infographics. The deck also inventories current multimodal tools (text, audio, image, video) and introduces practical pipelines, for instance, transforming recorded interviews into analyzable text and word-cloud summaries with GenAI assistance.

Ethics and transparency are treated as first-class concerns rather than afterthoughts. The talk operationalizes four principles for responsible use: Reliability, Honesty, Respect, and Accountability, and aligns them with actionable practices, such as, disclose tool use and methods; verify and reproduce claims; protect privacy and intellectual property; and maintain human agency and oversight. In literature workflows, the session recommends pairing general LLMs (for example, ChatGPT) with research-oriented tools (for example, Elicit, Consensus, SciSpace), while insisting on critical appraisal: do not accept outputs without checking consistency against the best available evidence, apply informal and formal logic, and verify compatibility with prior knowledge.

The conclusions balance opportunity and caution. GenAI demonstrably increases efficiency and expands the researcher’s toolkit, yet current limitations, especially around data and provenance traceability, demand measured adoption, explicit acknowledgments, and rigorous review. Used wisely, GenAI automates the repetitive and accelerates exploration, freeing researchers to focus on creativity, judgment, and intellectual autonomy, without displacing the essential human capacities that make research scientific.