Research Quality Is Under Threat from Speed, Fraud and AI

Reliable research requires detailed protocols, careful data collection, and thorough reporting of findings and limitations. Pressure to publish quickly can lead to rushed statistical coding, inadequate model checking, and, in some cases, manipulated or fabricated results. Rising publication volumes strain peer review, while artificial intelligence makes it easier to produce papers and code outside a researcher’s established expertise. This talk examines how these pressures undermine research quality and argues for changes that make published findings more reliable. Detailed records of a project’s provenance and stronger reproducibility checks, conducted automatically or by qualified researchers for high-impact papers, are proposed as ways to demonstrate and assess research quality.