AI in Science

Advancing Science with AI — and Advancing AI with Science

AI has become an integral part of how research is shared and improved—supporting everything from editorial workflows to personalized discovery. It supports:

  • Article recommendations and dynamic classification
  • Enhanced search and browse functionality
  • Image validation and annotation
  • Authorship and editorial matching
  • Content enrichment and structuring

Such integration enables better access to knowledge and streamlines the dissemination of trustworthy research. (Source: AI Ethics in Scholarly Communication).

Our members have long pioneered the responsible use of AI, applying it to content creation, workflow optimization, and the development of innovative tools.

To fulfill its promise, AI must be used responsibly and in alignment with the ethos of science. STM advocates for:

  • Accuracy and reliability: AI should operate on the final version of record (VoR) to ensure the most vetted, updated research is used.
  • Transparency and provenance: Systems must disclose sources, training data, and provide traceable references to maintain scholarly integrity. Many scientists are already concerned about downstream reuses of their works due to possible misrepresentations or misuse of their data for political gain.
  • Human oversight: Despite AI’s capabilities, human expertise remains essential to uphold the quality, trust, and accountability of scientific publishing.

Source: AI Ethics In Scholarly Communication

Also reference: Recommendations for a Classification of AI Use in Academic Manuscript Preparation

Despite all the potential gains, AI could also negatively affect knowledge production and dissemination in the research ecosystem if not handled carefully – and exacerbate an already pervasive blur between fact and fiction.

AI’s capabilities can amplify misinformation, especially when used without proper guardrails. The “hallucination” problem—where AI generates false or misleading content—poses a threat to scientific credibility and public trust.

Much like the spread of fake news through social media, scientific misinformation could erode societal trust in research and decision-making. Responsible stewardship is vital to counter these risks.

Society and policy-makers need to be able to trust scientific information to make evidence-based decisions, and researchers need to be able to drive innovation and discoveries that are so central to competitiveness and other societal benefits.

Recognizing the dual nature of AI, STM is leading initiatives that use AI to protect research integrity. The STM Integrity Hub exemplifies this approach, offering:

  • A shared, cloud-based infrastructure for detecting integrity issues
  • Integration with trusted tools like Springer Nature’s AI-powered text detection system
  • A human-in-the-loop model to ensure editorial discretion and accountability

This balanced approach ensures AI supports—not replaces—the essential work of human reviewers and editors.

LEARN MORE

Explore the STM Integrity Hub | A unified approach to safeguard research integrity

LEARN MORE

The latest AI news from STM

VIEW ALL NEWS

Just launched: the STM AI Portal

AI & Trusted Research   How AI shapes — and is shaped by — the academic record  From accelerating drug development to enabling green technologies, AI holds extraordinary promise for science — but it also introduces real risks, from AI-generated misinformation to large-scale manipulation of the academic record. As the pace of change accelerates, publishers and…

LEARN MORE

Cloudflare adds to their AI toolkit

The CDN provider has recently announced new tools and services for the era of AI. The two latest ones are their Content Signals Policy, which makes robots.txt indications more specific, and responsible AI bot principles. These may help bridge the gap between current practices and evolving legislation by providing effective tools to manage AI bots.

LEARN MORE

Upcoming EU initiatives

We’re expecting a busy autumn: after conducting consultations over the summer, the EU Commission will release a strategy for AI in science on 7 October and a Data Union Strategy at the end of October. A consultation was just launched on a Digital Omnibus, a package to review and streamline digital legislation to alleviate compliance burdens…

LEARN MORE

Danish working group on copyright and AI publishes recommendations

The Danish Ministry for Culture tasked a dedicated working group to explore questions about the interplay between AI and copyright. They delivered a very good set of recommendations, including recommendations on effective transparency in training data. EU countries will continue debating this topic into the fall, exploring potential solutions to favour AI licensing.

LEARN MORE