AI evidence extraction

Adoption of generative AI at the Italian institute for astrophysics: a survey on usage trends, ethics, literacy, and Shadow IT

Alessandro Cabras, Monica Marra, Giulio Capasso, Amedeo Petrella, A. Balestra, Ugo Lo Cicero et al. · Frontiers in Astronomy and Space Sciences · 2026

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

9/10

Relevance

2/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.3389/fspas.2026.1829012

Methodology & findings

Study design

Anonymous online survey administered over approximately 3 weeks (17 April to 9 May 2025) to INAF personnel across all structures and professional areas.

Sample

N = 371, 8 groups

Primary method

Descriptive statistics (percentages, frequencies, means, medians); Pearson correlation analysis; K-Means clustering (unsupervised machine learning); Elbow Method for optimal cluster determination; Silhouette score analysis (validation metric ranging from -1 to +1); Zero-mean and unit-variance scaling (StandardScaler); Pairwise deletion for handling missing data; Inductive thematic analysis for qualitative data (Italian-to-English translation); Heatmap visualization of frequency distributions; Boxplot analysis with median, mean, IQR, and whiskers; Radar charts for user persona visualization; Stacked bar charts for categorical distributions

Main result

The survey reveals widespread integration of Generative AI tools within INAF, with "50% (N = 186) of respondents reported using GenAI tools at least once a week, with a 17% (N = 63) reporting daily use." Notably, "there is a clear correlation between professional role and usage intensity," with early-career researchers showing the highest adoption. The analysis identified three distinct user personas: Enthusiasts (18.6%), Pragmatists (49.2%), and Skeptics (32.2%), with a critical finding that "Enthusiasts and Pragmatists overlap on usage and satisfaction, identifying a large productive workforce that diverges solely on the 'Personal Funds' axis, highlighting the demand for institutional licenses." A statistically significant positive correlation was found between AI literacy and satisfaction (r = 0.35, p < 0.001).

Reports effect sizes and confidence intervals.

Research paradigm

Mixed methods (quantitative survey with qualitative thematic analysis)

Author conclusions

The authors conclude: "This study provides the first quantitative mapping of Generative AI adoption within the Italian astrophysical community. The picture that emerges is one of a 'pragmatic revolution': INAF personnel are not waiting for centralized guidance or official infrastructure to integrate AI into their workflows. They are using it to write code faster, to bridge the linguistic gap in English writing, and to navigate administrative complexities." They identify three main institutional directives: "1. Recognize the 'Shadow Workforce'...2. Invest in Literacy, Not Just Policy...3. Differentiate Governance." The authors emphasize that "the challenge for research institutions is no longer whether to adopt AI, but how to proactively govern and support its use to ensure it remains a force for equity, security, and scientific excellence."

Risk of bias

Self-selection bias: respondents likely more interested in AI topics than non-respondents; Underrepresentation of early-career scientists (3.7% sample vs 27.3% population); Missing data primarily from survey branching logic and optional open-ended fields; Potential response bias: willingness to disclose AI usage may vary by role and ethical stance; Self-selection bias: respondents likely already interested in AI topics; Sampling bias: early-career scientists underrepresented (3.7% vs. 27.3% of population); Small sample size for early-career group (N=13) limiting generalizability; Survey completion affected by branching logic creating missing data patterns; Potential social desirability bias in self-reported AI literacy and usage; Under-representation of early-career scientists (3.7% vs. 27.3% of population); Sampling bias: 20% penetration rate may not fully represent all staff perspectives

Limitations

The authors acknowledge: "online surveys inherently suffer from self-selection bias
it is likely that respondents were already interested in AI topics
Therefore, the adoption rates might over-represent the 'tech-savvy' portion of the institute." Additionally, "early-career scientists (3.7% of our sample vs
27.3% of the population) are underrepresented
Consequently, subgroup percentages regarding this specific demographic...must be interpreted with caution due to the small sample size (N = 13), and should be viewed as exploratory indications rather than robust population-level findings." However, the authors note that "given the high overall response rate (20% of the workforce) and the broad coverage of all Research Groups (RSN) among permanent and technical staff, these trends provide a robust baseline for understanding the institute's core operational workflows."

Open questions raised

Prior research relied primarily on qualitative analysis of restricted working groups or limited survey samples
Lack of statistically representative empirical data on astronomical community integration of AI tools into daily workflows
Need for studies focused on specificities of national research infrastructures in astrophysics beyond German institutions
Emerging ethical concerns regarding use of AI in science evaluation (peer review, quality assessment)
Need for institutional reflection on ethical issues specific to national research bodies
Gap between transparent AI usage policies and actual disclosure rates in scientific publications

Data: Not explicitly mentioned as publicly available in the paperCode: Not explicitly mentioned in the paperExtracted from: pdfAgreement 56%

Explore related topics

Institutional Policy AI Governance Survey