Springer Nature has withdrawn a 2025 meta‑analysis that asserted OpenAI’s ChatGPT dramatically improves learning outcomes. The publisher cited analytical discrepancies and a lack of confidence in the paper’s conclusions after the article amassed hundreds of citations and widespread social‑media attention. Critics say the study mixed low‑quality research and rushed publication, raising doubts about its claims of large positive effects on performance, perception and higher‑order thinking. Read more →