AI Tools and Academic Productivity: My Erasmus+ Research Project at IPCB
During my Erasmus+ exchange at the Instituto Politécnico de Castelo Branco (IPCB), supervised by Prof. Sara Nunes, I designed and ran a full quantitative study on how AI tools such as ChatGPT influence students' academic productivity, understanding, and efficiency from survey design to SPSS analysis.
Why I chose this topic during my Erasmus+ semester
As part of my Erasmus+ exchange program at IPCB in Portugal, the Strategy and Competitiveness course required an original piece of quantitative research. I chose to investigate a topic that felt directly relevant to my own daily life as a student: the impact of AI tools on academic productivity.
Under the academic supervision of Prof. Sara Nunes, I built a complete research pipeline from designing the questionnaire, to collecting responses, cleaning the dataset in SPSS, and running statistical tests to answer a simple but timely question: does using AI tools like ChatGPT actually help students learn faster and better, or is it just a convenient shortcut?
Research objectives and questions
The study set out to explore whether the time students spend using AI tools is connected to how productive, efficient, and confident they feel in their academic work and whether that relationship depends on gender, field of study, or year of study.
Research questions
- RQ1: Does the amount of time spent using AI tools influence students' academic productivity?
- RQ2: Do students in different fields of study use AI tools at different frequencies?
- RQ3: Are there gender-based differences in AI tool usage?
- RQ4: Does the use of AI tools improve students' understanding of academic content?
Hypotheses tested
- H1: More time spent on AI tools is associated with higher academic productivity.
- H2: Usage frequency differs across fields of study.
- H3: There is a gender difference in AI tool usage.
- H4: AI tool usage positively influences understanding of academic content.
I wanted rigorous, testable evidence not opinions on how AI tools are actually reshaping student behavior.
Methodology: survey, sample, and SPSS pipeline
The survey was distributed to university students across multiple fields of study and covered three areas: weekly hours spent on AI tools, perceived improvements in academic productivity, and basic demographics (gender, field and year of study).
The dataset was cleaned and processed in SPSS. Out of 103 collected responses, one was removed due to missing data, leaving a final valid sample of 102 students. Text variables such as gender and field of study were recoded into numeric categories to run comparative tests (t-test, ANOVA, Chi-square) and correlation/regression analyses.
The project in numbers
Who answered the survey?
The final sample was fairly balanced between 46 women (45.1%) and 56 men (54.9%), mostly in their twenties, though ages ranged up to 50. Respondents were heavily concentrated in Business (56.9%), followed by Arts / Humanities / Languages (17.6%), with smaller shares from Social Sciences, Computer Science, Science, and Engineering.
In terms of study level, most respondents were in Master's Year 1 (31.3%) and Bachelor's Year 3 (25.3%), giving a good mix of early and advanced-stage students.
Among AI tools, ChatGPT dominated (67.6%), followed by Gemini (15.7%) and Claude (6.9%). Most students said they use AI mainly to search for information (71.6%) or summarize texts (22.1%) — rather than to write assignments directly.
What the statistical tests actually showed
Beyond the descriptive numbers, several statistical tests were run in SPSS to check whether these patterns were meaningful or just noise.
| Test | Question tested | Result |
|---|---|---|
| Pearson correlation | AI hours ↔ perceived quality improvement | r = 0.284, p = 0.005 (significant) |
| Pearson correlation | AI hours ↔ better understanding of content | r = 0.374, p = 0.000 (significant) |
| Independent t-test | AI usage frequency: women vs. men | t = -0.013, p = 0.990 (not significant) |
| One-way ANOVA | AI usage hours across fields of study | F = 1.597, p = 0.157 (not significant) |
| Chi-square | Field of study × AI usage frequency | χ² = 32.450, p = 0.019 (significant, cautious interpretation) |
| Linear regression | AI hours → perceived time saved | B = 0.074, p = 0.000 (significant predictor) |
In plain terms: gender didn't matter men and women use AI tools at almost identical frequencies. The field of study didn't strongly matter either for how many hours students spend on AI. What did matter was the amount of usage itself: the more hours students spent on AI tools, the more they reported saving time, improving the quality of their work, and understanding academic content better.
Discussion: what this means, and where it falls short
These findings echo recent literature suggesting AI tools are increasingly used as active learning partners rather than simple search engines while also showing, unlike some prior studies, that usage doesn't vary much by academic discipline in this sample.
Limitations I want to be upfront about
- Disciplinary bias: the sample skews heavily toward Business and Arts students; Engineering and Sciences are underrepresented.
- Tool bias: with ChatGPT dominating usage, the results mostly reflect one tool's impact.
- Modest sample size: 102 responses meet the course's minimum threshold but stay below the ideal target of 150.
- Self-reported data: answers rely on students' own perception, which can be affected by social desirability bias.
What this Erasmus+ experience gave me
Beyond the statistics, this project was my first time running a full research study end-to-end: writing a questionnaire people actually wanted to answer, cleaning messy real-world data, and defending statistical choices in front of an academic supervisor who pushed me to justify every test I ran.
Working with Prof. Sara Nunes at IPCB gave me a much sharper sense of how research rigor and business strategy intersect — a perspective I'm now carrying directly into my data and automation projects.
Curious about the full methodology or the complete dataset?
I'm happy to share the detailed report, the SPSS outputs, or talk through the research design — whether you're a fellow student, a researcher, or just curious about AI's real impact on learning.
Get in touch🏁 Conclusion
This Erasmus+ research project set out to answer a simple question does using AI tools actually help students perform better academically? and the data says: yes, but modestly, and mostly around time savings and comprehension rather than a dramatic leap in work quality.
Gender and field of study turned out to matter far less than the sheer amount of time students invest in these tools. For ChatGPT in particular, the picture that emerges is one of a fast, flexible research assistant — not yet a replacement for deep academic work, but clearly reshaping how students study.
Explore more of my projects
Beyond this Erasmus+ research project, I also build data and automation tools across Excel, Power BI, and Python — feel free to browse.
Frequently Asked Questions
What was this Erasmus+ project about?
It was a quantitative research project completed during an Erasmus+ exchange at IPCB (Portugal) for the "Strategy and Competitiveness" course, studying how AI tool usage relates to students' academic productivity and understanding.
Who supervised the research?
The project was supervised by Prof. Sara Nunes at the Instituto Politécnico de Castelo Branco (IPCB) as part of the Erasmus+ exchange program.
How many students took part in the survey?
A total of 103 responses were collected; after removing one incomplete entry, the final valid sample used for analysis was 102 students from various fields of study.
Which AI tool did students use the most?
ChatGPT was by far the most used tool, chosen by 67.6% of respondents, followed by Gemini (15.7%) and Claude (6.9%).
Does using AI tools actually improve academic performance?
The data shows a positive and statistically significant relationship between AI usage hours and perceived improvements in work quality (r = 0.284, p = 0.005) and understanding of academic content (r = 0.374, p = 0.000), though the effect size is moderate rather than dramatic.
Does gender or field of study affect AI tool usage?
No significant gender difference was found (p = 0.990). Field of study also showed no significant difference in weekly usage hours (ANOVA, p = 0.157), though a Chi-square test suggested a possible link with usage frequency that should be interpreted cautiously due to small subgroup sizes.
What statistical software and tests were used?
All data cleaning and analysis were performed in SPSS, using Pearson correlation, an independent-samples t-test, one-way ANOVA, Chi-square tests, and simple linear regression.
Want to discuss this research or my Erasmus+ experience?
Whether you're working on a similar research topic, curious about studying at IPCB, or just want to talk about AI in education I'd love to hear from you.