AI is both speeding up scientific discovery and quietly threatening the credibility of research findings, Saskia Steinacker, Chief Digital Officer at Springer Nature, shows in her DLD26 presentation.
“The question is, how can we keep science both trustworthy and fast to deliver solutions?”, she says. “And what role can AI play in this?”
On trustworthiness, she warns that “generative AI makes it easier now to produce very convincing but fraudulent scientific papers”, complete with fabricated data and images. “This can come from individuals”, she says, “but there are also large organizations who actually make this a business.”
With 2.3 million submissions in 2024 – “one every 10 seconds” – Springer Nature deploys AI tools like a nonsense text detector and an irrelevant reference checker to flag problems for editors.
On speed, Steinacker notes that “only 45% of all researchers say they have enough time for the core research.” This is where AI-powered literature overviews and writing tools can play a positive role. “That’s something that helps researchers for a first draft”, she says. “They can structure the manuscript. It helps with the clarity. But also think about non-native speakers.”
Watch the video to explore this talk in detail.
