Home > Useful contents > AI in research: a help or a hidden risk?
AI in research: a help or a hidden risk?
Artificial intelligence promises to make science faster and more efficient. But are we sure that is really the case? According to Holden Thorp, editor-in-chief of Science, the rise in articles written or supported by AI is creating new problems: errors that are difficult to spot, fabricated citations, texts that appear convincing but are scientifically weak, and a growing workload for reviewers and editors as they carry out increasingly thorough checks and verifications.
For university students, this presents a significant challenge: passively relying on AI can compromise the quality of research and the ability to develop genuine critical thinking. In a context where automatically generated information is increasing rapidly, knowing how to verify sources, interpret data and recognise errors is becoming an essential skill for every future professional and researcher.
A question to ponder: if a text appears perfect but contains hidden errors, are we still able to distinguish it from sound science?
Holden Thorp (2026) AI in scientific publishing: Slower, worse, and more expensive. Science 393: 225-225(2026). https://www.science.org/doi/10.1126/science.aek5570
