Javier García Manglano: "The process is what really makes us grow as researchers: reading, writing, thinking."

How can artificial intelligence tools transform the way we conduct research? From reviewing literature and analyzing data to writing code and visualizing results, AI is part of the daily routine for many university employees.

Javier García Manglano, researcher of group Youth in transitiongroup at the Institute for Culture and Society (ICS), explains what his team is currently studying, how it uses AI tools to advance its work, and what opportunities and risks this new stage poses for research

What research currently involved in?
We have two areas of specialization: young people's transitions to adulthood, and the role of technology in young people's interpersonal relationships. So far in 2026, we have published articles on young people's perception of adult life (which is somewhat negative), on the relationship between the movies and series one watches as a teenager and the risk of consuming pornography as an adult (there is a steep slope), and on improving the way we measure cell phone dependence (with information on the context in which it is used).

What AI tools do you use for your research?
Since last summer, I've had a subscription to Gemini Pro and am looking for different ways to get the most out of it. But my tool is NotebookLM, also from Google; I find that it allows you to limit the risk of hallucinations without losing the potential of AI. For those who don't know it, it works with notebooks, each of which contains sources entered by the researcher. Once the notebook is created, you can ask it questions, request a concept map, create study cards, or make an infographic or a summary video or audio format... but only from your sources. One use I'm enjoying is talking to "experts": I put in everything researcher reference letter researcher has published reference letter can ask questions to simulate a conversation with that person... Imagine being able to talk to the people who know the most in your area!

Could you specify how AI has helped you with your research?
Perhaps the most efficient use I make of AI is to save time writing statistical code. I have always used software such as Stata, but I have knowledge limited knowledge of other languages such as R. I can ask Gemini to write code for me in either language (or both, for comparison), or to find an error or bug in my code... and it does so much faster than any human. Of course, I can use it this way because I have spent almost two decades writing code myself, so I can review what the AI creates, make changes or ask it to make changes, and thus control the process and ensure the methodological rigor of what I produce.

How did you work before the advent of AI?
Another aspect I've always wanted to devote time to but never got around to is the visualization of results: a good graph can be the best way to explain or disseminate scientific results. And what was once unattainable, I can now ask AI to do, without risk: because I understand the results, I can try out and ask for different ways of visualizing them and choose the one that convinces me the most.

When did you realize that AI was a tool would facilitate your work and, therefore, was here to stay?
In the last year. The truth is that I had hardly used it in previous courses, and on occasions, the experience had been frustrating. But the latest paid models (the free ones are at least a year behind) are much more reliable and capable. Perhaps I have also learned to use them better, more as assistants in areas I already master and not for tasks where I don't have enough knowledge judge the result . I think it's a combination of both factors.

Could you do research again without AI?
Of course I could. Maybe it would be slower, but perhaps I would also proceed more cautiously and make better use of staff; sometimes, I worry about trying to take on too much and losing patience for the tasks that "build" you as researcher. In other words, I think there is a real danger of separating the result process. The result what we are evaluated on: basically, publications. The process is what really makes us grow as researchers: reading, writing, thinking. There is a risk of entering a degree program frenetic about the former, and even more so when everyone is taking advantage of these tools, that we end up neglecting the latter. That is, personally, the danger I fear the most.

On ethical and manager use: how can AI be used to increase research efficiency research sacrificing authorship, transparency, and ethics?
I would like to think that the answer to this question is exemplified in the previous ones, because that is what I try to do in my research work. But there are uncomfortable questions that we will have to ask ourselves sooner or later. Should we really spend time on tasks that AI does much faster than we do, with result acceptable result ? Or should we delegate those tasks to AI and devote ourselves to what adds human value? How much of what we were doing until recently can we stop doing without prejudice to our development research work?