Science, technology and society articles from across Nature Portfolio

Biases in artificial intelligence models have been studied predominantly through Western lenses, overlooking South Asia’s unique contexts of caste, religion, colourism and representation. This Comment highlights region-specific biases in language and vision models and calls for fairness frameworks grounded in South Asian realities.

Generative artificial intelligence is transforming the way scholars draft, revise, and publish. Yet, academia lacks a systematic way to measure these shifts and risks relying on anecdotal evidence in evaluating whether AI elevates or erodes scholarly standards. This Comment draws on a pre-registered, three-day field experiment that addressed this lack of measurement by pairing twenty-two early-career researchers with and without AI tools to improve scholarly manuscripts for journal submission. However, the AI models used in the field experiment are already outdated and outperformed by more powerful reasoning models, situating the results as a snapshot in time. This Comment calls for recurring events with a similar set of evaluation criteria to combine the results in a publicly available dataset. Monitoring the quality of researcher-AI collaboration is necessary if academia wants to keep track of AI’s rapid impact on research practice.

Marta Koch explains how stakeholder asset-mapping can help identify emerging climate action technology solutions in the Arctic.

Most policy proposals aimed at managing the risks of artificial intelligence (AI)-enabled weapons rely heavily on meaningful human control or appropriate human judgment for risk mitigation. This Comment argues that there are various ways humans can exert such control over AI, and that developing a careful taxonomy of these is necessary for building actionable risk-mitigation policies for warfighting AI.

The organizers reflect on how a multi-year, multi-country benchmark aligned AI research in road damage detection with practical and regional constraints, steering it towards deployment relevance.