Peer-reviewed sources, verifiable citations: What convinced one PhD researcher to try AI

Created with researchers in mind

As a time-pressured PhD researcher in his third year at Germany’s RWTH Aachen University, Paul Preuschoff understands why some colleagues turn to big tech AI tools for support.

“There have been times when deadlines were approaching that I wasn’t sure I could resist,” he admits, “but I have never used AI tools like ChatGPT — I just distrust them too much.” One of the factors driving this wariness is the number of times Paul has seen it “go wrong” with his own students in the university’s Human Computer Interaction (HCI) department: “I don't want to go down that rabbit hole,” he explains.

Recently, however, Paul was invited to provide feedback on LeapSpace, Elsevier’s research-grade AI workspace. Launched in January 2026, LeapSpace brings together peer-reviewed abstracts and full-text content from leading publishers and scholarly societies. Outputs include citations to support traceability. This, along with LeapSpace’s scholarly foundations, have given Paul the confidence to start exploring AI.

“The fact that LeapSpace uses only peer-reviewed sources is a huge thing for me — I don't want to get claims that are taken from newspaper articles or popular media,” he explains. “LeapSpace is not a general tool marketed for a broad population, it is created with researchers in mind, which means I have more trust in it. And being able to check the references makes me feel I am still in control of what I do with this information.”

Driving progress with Deep Research

The LeapSpace feature Paul uses the most is Deep Research mode, which leverages a multi-agent AI system to explore a topic from multiple perspectives. Findings are shared in a comprehensive report that is designed to show the first principles thinking approach employed; the evidence uncovered, along with confidence signals and trust markers. For Paul, this offers the opportunity to:

Validate hypotheses

Paul explains: “I always start with some wild idea inspired by research or my everyday experiences and then I want to know has it been done already? Deep Research helps me refine where I want to go in my research by making it easier for me to choose from multiple options the one that makes sense. It also makes it easier to learn outside of my domain.”

Move forward, faster

For his PhD topic, Paul is looking at how AI can enrich creative processes without undermining people’s sense of ownership and joy. “But because AI is a fast-moving field, it can be really hard to keep up to date, especially with new topics,” he says.

For Paul, the Deep Research report offers an easy — and time-saving — way to get to grips with an unfamiliar topic. One benefit is the intuitive search: “You always have this fear that if you use the wrong terms you may miss that core paper that could open up the whole space,” Paul says. “Deep Research has repeatedly shown that if I enter some vaguely related terms, it will find the ones primarily used in literature. And if it doesn’t find anything, while I still double check that with conventional search, it does give me confidence that there probably isn’t anything to find.”

Another benefit is the breadth of insights in the report. He points to a recent query he posed about how AI can be linked to tabletop RPGs (role-playing games). “The Deep Research report was generated typically within minutes and after reading it I felt like I knew roughly what's going on in the topic — getting a broad overview like that would normally take me a day or more.” He adds: “Given that I don’t actually have any spare days, Deep Research propels me to a point in my reading I wouldn’t reach otherwise.”

Span disciplinary borders

Another plus point for Paul is that LeapSpace and Deep Research look beyond his own domain for sources — results are found and ranked based on their relevancy to the query. “I really feel we could profit from looking at research from more perspectives, but the moment I move outside computer science it is easy to get lost — knowing where to start looking and I understanding the vocabulary. That makes it hard to say whether something is relevant.”

“But with LeapSpace I can ask cross-disciplinary questions. And even when my questions are computer-focused, it may come back with results beyond my area of expertise. For example, when I asked about RPGs, it went to the gaming side of research and that offered a perspective I hadn’t really thought about before — apparently sometimes just using a computer to have fun is enough!”

Paul believes Elsevier’s goal to make LeapSpace publisher neutral will further expand cross-disciplinary coverage. Agreements have already been signed with Emerald Publishingopens in new tab/window, IOP Publishingopens in new tab/window, NEJM Groupopens in new tab/window, Oxford University Pressopens in new tab/window and Sageopens in new tab/window, to index and make searchable a selection of their full-text journal content in LeapSpace, with more publishers in the pipeline. LeapSpace is designed to include a link to the full text of the article on the publisher’s platform for their subscribed users to access whenever their content is referenced.

Finding potential reviewers and collaborators

Paul is also looking forward to experimenting with LeapSpace’s vector-based Author Search tool. Because it doesn’t rely on exact keyword matches, it can find researchers in adjacent or complementary fields. This makes it a valuable tool for cross-disciplinary research.

It also has the potential to help Paul with a problem that many researchers struggle with — identifying researchers to recommend as manuscript reviewers. “We try to think about it before the submission deadline, but then things can get so stressful we don’t get around to it. To be able to just describe what we are investigating and get a list of relevant people is promising.”

He adds: “As a PhD student in my first years, it would also be very interesting to connect to people in the same position; however, as these people haven’t published so much, they don’t always appear in search results.”

Paul also likes the fact that you can instruct LeapSpace to filter by h-index or number of publications. “A computer science colleague of mine has to find brain computer interface specialists that typically focus on medicine so this could really help there.”

He even sees its potential to identify experts in complementary fields within his own institution. “Our university has had an Institute for Man-Machine Interaction focused on electrical engineering for the last 20 years — I only found out about it last week!”

Building to meet real research needs

Paul’s feedback also included some improvement ideas for the LeapSpace development team. While he finds the structure and content of Deep Research’s reports helpful, he would like to see some changes to their text. “I do see some AI language used and that’s something I don't want in my scientific thinking. For example, I don’t want to be told something is ‘rich ground.’ I want to see the reasons why that claim is being made and then conclude that for myself.”

And as Deep Research reports can take several minutes to run, he would like the option to stop them midway if he can see a way to improve the prompt. The final item on his wish list is an integration with reference management systems. Since April 30th this year, users can now export references from LeapSpace in CSV, RIS, BibTeX and plain text.