Academics divided on AI for research writing, and other digital technology stories you need to know

A Nature survey on whether researchers believe AI should be allowed to write science papers has found “contrasting views”, the publication has said.

In the poll of 5,000 academics from across the globe, more than 90% of respondents thought it was acceptable to use generative AI for editing or translation purposes. But they differed on whether the use of AI should be disclosed and the level of disclosure needed.

Writing all or part of a paper with AI was felt to be ethically acceptable by 65% of those polled, while a third were against it.

More than 60% of respondents, meanwhile, felt it was not appropriate to use AI to conduct an initial peer review of a research manuscript. But 57% said it was OK to use AI to assist in peer review by answering questions about a manuscript.

The number of papers with signs of AI use is rising rapidly, Nature says. Only a minority of researchers in the study said they had actually used generative AI to edit, translate or write their papers.

The United Kingdom and the European Union have revealed plans to “supercharge” artificial intelligence (AI) collaborations across Europe.

Building on the UK’s new deal with the EU, the partnership will invite UK public research organizations to apply to become the UK’s link to advanced super-computing facilities in Europe.

Greater collaboration will help address global challenges and support the development of advanced AI systems used in healthcare and energy, the UK government says.

Recent research from the World Economic Forum underscores AI’s potential impact. According to the white paper Blueprint for Intelligent Economies – AI Competitiveness through Regional Collaboration, technology is increasingly acknowledged as a transformative enabler, fundamentally changing industries, economies and societies worldwide.

AI could account for almost half of data centre power consumption by the end of this year, according to new estimates based on the power consumed by chips used to train and operate AI models.

Researchers and governments in the Global South are increasingly focusing on developing their own generative AI models, according to a report in Nature India. Instead of looking to simply scale up the technology, researchers across Africa, Asia and the Middle East are looking to “scale right” – developing models in local languages that reflect cultural nuances and social an economic realities.

Nearly half of US technology firms are already integrating or fully deploying agentic AI – autonomous systems that can make decisions and perform tasks without human intervention – according to a survey from EY. Half of those surveyed also said that more than 50% of their AI operations will work autonomously within the next 24 months.

A start-up is trialling a cloud-based software platform that helps healthcare providers manage disparate tools and masses of healthcare data in one place. Innovaccer hopes its Gravity platform will eventually improve patient care and make hospitals more efficient.

Google DeepMind says it has developed a general-purpose AI that can devise solutions to major mathematics and science problems. AlphaEvolve – described by the company as “a coding agent for scientific and algorithmic discovery” – combines large language models with algorithms that scrutinize suggestions to filter and improve solutions.

Six trains in Scotland have been fitted with satellite technology from Starlink to supply passengers with WiFi. The six-month trial is an onboard passenger WiFi first for the UK, according to the train operator, ScotRail.

Using digital technology is associated with a 58% reduced risk of cognitive decline in people middle-aged and older, a new study says. The research, published in Nature Human Behaviour, says that engaging with devices including smartphones, tablets and computers could play a role in preserving brain function rather than worsening it.

Roughly 90% of businesses see agentic AI as a potential source of competitive advantage, gained through efficiency, enhanced decision making or scalability, according to recent research. But as the prevalence and impact of AI agents rise, the importance of understanding and selecting the right model will rise too.

In a world in which nearly 2.6 billion people remain offline, the datasets underpinning AI systems don’t yet reflect the full diversity of human experience. Modernizing data infrastructure is essential for addressing this and scaling AI responsibly and securely – and inclusive AI requires collaboration, governance and a long-term commitment to ethical design.

Research shows the majority of people find the online buying experience frustrating. But AI agents are about to change all of that – and, according to Benjamin Wiener of technology solutions company Cognizant, the way customers engage with brands will never be the same.