ResearchRabbit update aims to keep students curious

ResearchRabbit has introduced product updates aimed at helping graduate students explore academic literature more freely while keeping sight of their original research question. The changes come as universities and researchers rethink how discovery tools should support inquiry in the age of generative artificial intelligence.

The update includes a new feature called Analyse, designed to prompt users to look beyond the immediate boundaries of their topic and draw in ideas from adjacent areas. The wider changes are intended to support exploratory searching while reducing the risk that researchers lose sight of their starting point.

Wellington-based ResearchRabbit focuses on literature discovery through citation networks rather than generative AI. Its platform maps more than 320 million papers and lets researchers navigate links between publications based on their own choice of source material.

The launch reflects a wider debate in academia over how research habits are changing as tools such as ChatGPT become more common in higher education and scientific work. The AI for Science 2026 report from Nature Research Intelligence, based on responses from more than 10,000 researchers worldwide, found that 44% use AI for information gathering every time or most times.

The report identified general-purpose models as the most widely used AI tools among researchers. That shift has intensified discussion over whether academic training should focus less on retrieving answers and more on framing worthwhile questions.

Auckland University of Technology Associate Professor Mahsa McCauley said literature reviews should remain a place for exploration rather than become a procedural exercise.

"The literature review is where it starts. It shouldn't be a box-ticking summary of what's already known. It's the place to wander, follow the threads that surprise you, and come back with a sharper question than the one you started with. Tools that let students explore, without losing their way, make that wandering productive rather than risky," said Mahsa McCauley, Associate Professor and Head of Computer and Information Sciences at Auckland University of Technology.

McCauley also argued that the spread of AI is changing what universities should value in research training.

"For a long time we trained students to be answer hunters," McCauley said.

"Now AI can produce answers in seconds, so the real skill is knowing which questions are worth asking. I want students to become question farmers, people who cultivate curiosity on purpose. The future of work won't be about finding answers but about finding better problems to solve, and that comes from human judgement, care, and curiosity," she said.

Curiosity debate

ResearchRabbit framed its latest changes within that broader shift, arguing that research software should preserve active user involvement rather than automate conclusions. It does not currently use generative AI, instead relying on algorithms built on citation networks seeded by papers selected by the researcher.

That distinction is central to how the company sees its role as AI tools gain ground in academia. Rather than produce responses, it aims to guide users through connections in the literature and let them pursue unexpected lines of inquiry.

"We aren't building a tool that generates answers. Answers are a termination point; rushing to answers endangers creativity, curiosity, and deep exploration. We know our users will move through what we've built in their own unique way. The power of ResearchRabbit is in combining a huge database of more than 320 million papers with a unique way of exploring this information," said Digl Dixon, Head of Product at ResearchRabbit.

The emphasis on different styles of inquiry also echoes academic work on how people pursue knowledge. A 2024 study in Science Advances by researchers from the University of Pennsylvania and the Wikimedia Foundation examined nearly half a million Wikipedia readers across 50 countries and territories and identified several patterns of curiosity, including users who search broadly, those who pursue a specific answer, and those who move between unrelated areas.

For ResearchRabbit, the argument is that software should support those different patterns without pushing users too quickly towards a fixed outcome. The latest update, including prompts to scan the edges of a field, is intended to create more room for that process.

"We are constantly working to retain cognitive involvement for our users and support exploration. With this update we are making it even easier to fall down rabbit-holes, while encouraging users to look to the edge of their research and bring in new thinking. That is often where cross-pollination of ideas happens, and it can lead to unique scientific breakthroughs," Dixon said.

The platform has been in use since 2021, and 4.5 million people have used it.