AI's growing role in scientific peer review

Large language models can enhance scientific peer review by identifying objective errors and inconsistencies, improving draft quality, and alleviating reviewer workload. However, AI is less effective at subjective judgments, such as assessing novelty or significance. Human oversight remains essential, with transparency about AI involvement and accountability for final decisions. AI's role in science is expected to expand.

There is tremendous interest in using AI, especially language models, to support research and peer review and speed up the scientific process. A key advantage is that AI can act like a rapid, always-available critic, a sort of pre-submission review process—before scientists officially send in a paper for publication. AI can be quite good at assessing drafts for gaps and limitations, so researchers can preemptively address them. This can improve the quality of first drafts submitted for publication and reduce the back-and-forth later. And on the reviewer side, the pressure is real: As submissions grow, human reviewers are very much overburdened, which can lead to lower-quality reviews and frustration for authors.

IN their tests researchers found that  besides spotting gaps and limitations, AI can be quite good at the more objective, verifiable aspects of review. 

 AI is strongest on objective, checkable inconsistencies and technical issues and weaker on subjective judgments about the novelty or significance of the research.

The researchers say that AI should support and inform—not fully replace—human decision-making.

A human, or team of humans, must make the final editorial decisions and scientists must stand behind the work. AI can offer comments on early drafts, point out omissions, and suggest improvements in the writing and the research—but the scientists must remain accountable for incorporating and synthesizing feedback from the AI and human reviewers. 

Scientists have to be up-front about how and where AI has assisted in the research itself and in the writing and review of the papers. They should acknowledge exactly how AI was involved and what tools were used in the paper. It comes down to accountability and a clear chain of responsibility and that final decisions are still made by humans.

Following on this work, many conferences and journals are now exploring using LLMs to assist the review process.

Nitya Thakkar et al, A large-scale randomized study of large language model feedback in peer review, Nature Machine Intelligence (2026). DOI: 10.1038/s42256-026-01188-x

In one recent large-scale randomized experiment, the results of which were published in Nature Machine Intelligence, researchers provided AI assistance to human reviewers on roughly 20,000 reviews to assess AI's impact on review quality. 

**

Premature and small births are linked to lifelong learning problems

Preterm birth and low birth weight are consistently associated with lower IQ and persistent educational disadvantages, particularly in mathematics, from early childhood into adulthood. The severity of these challenges increases with earlier gestational age and lower birth weight, underscoring the importance of early identification and ongoing support to improve long-term outcomes.
Being born early or at a lower weight is linked to lower IQ scores and poorer educational outcomes in school and beyond, according to a new study published in the journal JAMA Pediatrics.
In this research, known as an umbrella review, the team examined what previous studies had discovered about preterm birth and low birth weight and long-term development. This involved going back to the original numbers and recalculating the results using a single, consistent method to ensure accuracy. They looked at five different life stages, from babies under two years old to adults over 18.

This meta-analysis confirmed that both preterm birth and low birth weight are linked to disadvantages that persist over time. In particular, babies born before 28 weeks or weighing less than 1 kg at birth showed larger academic disadvantages on average than babies born at term with normal birth weight.

The most affected subject was math, with significant gaps in calculation and problem-solving skills. Stark differences were also seen in reading, comprehension, spelling and identifying words.

These challenges were often most visible during primary school and closed slightly during teenage years. However, some of these learning difficulties reappear once a person reaches adulthood, as the study authors note in their paper. "These disadvantages generally increased with earlier gestational age and lower birth weight. Although some associations appeared to attenuate during adolescence, evidence of persistent disadvantages into adulthood was observed for several outcomes."

The research team believes their findings show that the impact of being born early or much smaller than average can have lifelong consequences. For some, this may mean fewer job opportunities or earning lower salaries than their peers.

Mingzheng Hu et al, Cognitive and Educational Outcomes After Preterm Birth or Low Birth Weight, JAMA Pediatrics (2026). DOI: 10.1001/jamapediatrics.2026.0533

**

How time and space become one inside your brain—and what it means for Alzheimer's

If you develop Alzheimer's disease, you not only lose your sense of time, but you also lose your sense of place.

Neural circuits in the retrosplenial cortex process time and space using similar activity patterns, indicating these dimensions are integrated in the brain. This shared mechanism helps explain why both temporal and spatial orientation deteriorate together in Alzheimer's disease, highlighting the need to understand healthy episodic memory networks to address dementia.

All memories are made up of different components. You don't just remember what you had for dinner yesterday, but also the time and place. We often think of time and space as separate categories, a distinction created by philosophers and physicists that is incredibly practical for organizing our lives. But our brain cells don't see it that way.
These cells don't distinguish between a step forward in space or a second passing in time. Instead, they simply record a continuously changing stream of information from our senses, tracking events as they unfold. To the brain's internal network, time and place are effectively two sides of the same coin.
In Alzheimer's disease, it is therefore not surprising that both are affected; when the neural network is damaged, our sense of 'where' and 'when' begins to unravel together.
Remembering where, when and how something happened is called episodic memory. In your brain, billions of nerve cells form large networks, passing signals like a relay race to process information from your senses, the sounds, smells, and sights of your life.

We already know that cells which link memories to time and space are found in the hippocampus.
But this group of researchers had a theory that another area of the brain is also involved, namely the retrosplenial cortex. Located at the back of the cerebral cortex near the hippocampus, this area was previously only known for linking memories to place.
To test if this area also tracks time, the team designed a memory challenge for mice. The task required them to hold a specific odor in their "working memory" during a brief period. Their study is published in Cell Reports.
The most striking discovery was that the retrosplenial cortex uses the same "neural script" for both space and time. The researchers found that the sequence of neuronal activity in the retrosplenial cortex looks almost identical whether a mouse is physically running through a room or simply holding a memory in its mind for five seconds.
This discovery brings us back to the tragic reality of Alzheimer's disease, where those affected struggle to anchor themselves in both time and place. By showing that the brain uses the same "neural script" for both, this research explains why these two senses often fail together.
This work also challenges how we perceive the world around us. While we use the concepts of time and space to organize our lives, this distinction is largely a human construct. In fact, some modern theories in physics are moving away from using time and space as the fundamental building blocks of the universe. It appears the brain's internal wiring mirrors this deeper reality.

Anna Christina Garvert et al, Area-specific encoding of temporal information in the neocortex, Cell Reports (2025). DOI: 10.1016/j.celrep.2025.115363