When can we say AI made a scientific discovery?

Anthropic recently claimed that its AI agents made a scientific discovery by identifying a unique DNA pattern related to enzyme function. While the company framed this as a breakthrough, the announcement sparked backlash from the scientific community. Biologists argue that identifying patterns is merely a preliminary step, not a discovery in itself. Critics emphasize that true scientific progress relies on understanding biological mechanisms, not just data processing. Furthermore, a researcher claimed his team had already identified the pattern, raising questions about AI's originality and the transparency of training data. This incident highlights a growing tension between AI companies, which seek to market their systems as autonomous researchers, and the scientific community, which views AI as a tool rather than a peer. As AI models increasingly assist in complex research, the debate over what constitutes a genuine 'discovery' remains a contentious point in the evolution of scientific methodology.