In a recent article for Nature, Helena Kudiabor explores how AI can help scientists conduct research, while acknowledging some of its challenges. Kudiabor writes that some popular large language model (LLM)-based tools can be used to expedite literature reviews by providing useful insights for a high-level scan of a topic. Epidemiologist Razia Aliani discussed how the choice of AI tool depends on the task, noting their usefulness in reviewing large databases. However, other academics raised concerns about accuracy, particularly regarding ChatGPT’s tendency to “hallucinate” false information or invent references. Alec Thomas (University of Lausanne) discussed why he stopped using AI search tools, attributing this decision to the serious errors he encountered while using it.
