How researchers can use GenAI to get the info they need quicker than ever
15 de fevereiro de 2024
Por Ian Evans
ポルトガルのUniversity of Aveiroで博士課程に在籍するBruno Augusto氏（写真：Michael Russo氏）
After testing Scopus AI, an environmental engineering researcher talks about how it enables him to work faster, delve deeper and cross disciplines.
As a researcher in environmental engineering, Bruno Augusto(abre em uma nova guia/janela) has an acute understanding of the dangers of misinformation. So when he got involved in testing Scopus AI — Elsevier’s new generative AI search tool — reliability of information was high on the list of his priorities.
“It’s easy to check the references”
“Something that gives me a lot of confidence in this tool is that the results are referenced, and it’s easy to check the references,” said Bruno, a PhD student at the University of Aveiro(abre em uma nova guia/janela) in Portugal. “You still need to use some critical judgment when you’re reviewing the results, but as the tool gives you all the references it cites from, you can go through those, see if they fit what you need or if there’s a better result — and it makes the process much quicker.”
Launched in January, Scopus AI draws from metadata and abstracts of Scopus documents published since 2013. Advanced prompt engineering and curated recent data minimize risks of false AI-generated information and ensure responses are based on recent, trusted knowledge. The idea is to provide a trustworthy guide through the huge amount of knowledge found on Scopus, the world’s largest multidisciplinary abstract and citation database.
For Bruno, the biggest change is the speed at which he can get to the article reference he needs. He explained:
When you’re searching for a reference to go with a specific sentence in your paper, you used to have to do a keyword search and go through 10 papers to get what you need. Just this week, I used Scopus AI to search for the sentence I wanted a reference for. From those results, I had two or three papers I could look through, rather than going through the 10 or 20 papers that a key word search would show.
To illustrate the point, Bruno referred to a recent example in his own work.
Last week, I was searching for papers about urban areas being more vulnerable to air pollution and emissions. So I just searched directly on that exact term rather than searching for ‘emissions, air quality, urban areas.’ Scopus AI gave me a couple of quick citations. One of them wasn’t what I needed, but the other one was exactly what I was after. I checked the article to make sure it was a good reference, and then I was able to include it.
Bruno Augusto
The ability to reference claims in a piece of generative AI content is a crucial part of why Scopus AI works for researchers. Bruno explained: “Most of science is very context dependent, so if, for example, Scopus AI gives you a result that’s the opposite of what you were looking for, or what you expected, then you want to double check the references.”
In that sense, he said, generative AI remains a useful tool that speeds up processes for researchers rather than a replacement for any elements of the research process.
Not for the lazy
“It still requires critical thinking from the user,” Bruno said. “If you were inclined to be lazy and just assume everything is exactly correct without following up on the references, then you might run into problems. Again, that’s where the references are helpful — because you can see the context. So I think, as with any tool, there’s the need for a bit of user education and critical thinking about what you use and how you use it.”
Breadth and depth: Breaking down barriers between disciplines
Reflecting on the best moments of his time spent with Scopus AI, Bruno said it’s the sense of speed that leaps out to him the most: “It really facilitates the search process.” But he also highlighted the way it can break down barriers between disciplines by allowing you to search for answers using every-day language. As someone whose research often crosses disciplinary boundaries, that can be extremely useful, he explained:
One thing that’s a problem with science in general is that communication is often not great. Before you even start building your knowledge in an area, you have to know a lot of technical terms. But with Scopus AI, you can use more informal language when searching, and you get pretty much the same results as you do with the technical terms. I think that’s one of the great benefits of using the tool.
Bruno also noted that being able to drill down into the results was helpful, allowing you to follow up on an initial query. “The follow up questions were phenomenal,” he said. “You can get as deep into topics as you need.”
Suggestions for improvement
That said, Scopus AI is a new product and there is room for improvement. In particular, Bruno noted that while the focus on research since 2013 keeps the information up to date, it could be useful for Scopus AI to draw from a larger pool.
“I think having a toggle for the years of research would be interesting,” he said. “A lot of those older papers provided the foundation for later work, so it’s useful to have that in the parameters, as we have in regular Scopus searches. Similarly, it would be useful to toggle for citation number — that could make it even more useful.”
Continuing the journey with the research community
Commenting on Bruno’s experience, Maxim Khan(abre em uma nova guia/janela), Elsevier’s Senior VP for Analytics Products and Data Platform, said:
User feedback has been really important in bringing Scopus AI to life. Bruno is spot on - there is still a lot more we can do. It’s energizing to hear how Scopus AI is empowering researchers to work quickly and that the emphasis on reliable content is also recognized. Throughout the whole process of developing Scopus AI, it’s been fantastic to hear user responses and how much they’ve come to appreciate this new tool. I’m looking forward to continuing the journey with the research community.
Contribuidor
IE