📊 Study Overview
The ATIS Framework emerged from a rigorous three-round qualitative study designed to identify success factors for global AI projects by examining the interaction between technical triggers and cross-cultural dynamics.
"Data science is that like a lot of times, we don't know. It's different. We don't know the answers, right?"
— P5, Data Scientist (from interviews)
🔬 Research Methodology
The study employed a hybrid computational-qualitative approach across three stages:
📝
Stage 1
Exploratory Inductive
Manual thematic analysis establishing 35-subtheme taxonomy with "human-in-the-loop" grounding
🤖
Stage 2
Abductive Expansion
LLM-powered "Latent Semantic Scan" revealing emergent codes (stochasticity, autocatalysis)
📈
Stage 3
Confirmatory Deductive
6-phase hybrid validation with Python-based clustering and 44 seed papers
⏳ Research Timeline
March 2023
Round 1
5 expert interviews focused on AI lifecycle and organizational context
May 2023
Round 2
5 interviews testing emergent themes; abductive pivot to include organizational and leadership agency
March-April 2025
Round 3
38 large-scale interviews for computational validation (post-LLM deployment era)
👥 Expert Panel N=48
Maximum variation sampling across roles and geographies (Patton, 2015)
🌍 Geographic Representation
🇺🇸 USA
🇬🇧 UK
🇫🇷 France
🇧🇷 Brazil
🇮🇳 India
🇰🇪 Kenya
🇯🇵 Japan
🇲🇽 Mexico
🇸🇪 Sweden
🇳🇱 Netherlands
🇦🇪 UAE
🇨🇱 Chile
🇵🇪 Peru
🇳🇵 Nepal
🇹🇿 Tanzania
🇵🇰 Pakistan
🇨🇳 China
🇩🇪 Germany
📚 Related Publications
Explore additional research on AI, creativity, and cross-cultural dynamics:
LLM-Creativity: Measuring and Explaining Creativity in Large Language Models
October 2025 • Mixed-Methods • N=6,375
Investigates how theory-driven cultural persona prompts affect creative output in business problem-solving. Reveals criterion-dependent effectiveness and the trade-off between novelty and feasibility.
DOI: 10.5281/zenodo.17407392
View on Zenodo →
The LLM Team Composition Paradox: Why the Right Team Isn't Always Diverse
October 2025 • ARC-AGI Benchmark • OODA-Belbin Framework
Challenges the assumption that architectural complexity guarantees superior performance. Introduces a contingency model showing when single agents, homogeneous teams, or diverse teams perform best.
DOI: 10.5281/zenodo.17490945
View on Zenodo →
ATIS Framework: AI Project Management in Cross-Cultural Contexts
February 2026 • 48 Experts • 1,392 Units
The current research—introducing the ATIS Framework, Causal Chain Model, and Maturity Model for managing AI project friction across cultures.
DOI: 10.5281/zenodo.17425216
View on Zenodo →
Research theme: These three papers form a cohesive body of work on AI, culture, and human-AI collaboration—from understanding cultural prompts in LLMs, to optimizing AI team composition, to managing cross-cultural AI projects.
📡 Recruitment Channels
Participants were sourced from 7 prominent AI and Data Science communities on Slack, plus professional networks.
Data Science Salon (DSS)
Data Talks Club
TWIML Community
ODSC Global
Convergence
DRE Community
MLOps Community
Professional Networks
📚 Theoretical Foundation
The study's theoretical rubric was developed from seminal seed papers for each AI specificity:
- Autocatalysis: Su & Ayob (2025) - Artificial Intelligence in Project Success
- Stochastic Uncertainty: Fridgeirsson et al. (2021) - Effect of AI on Project Management
- Process Disruption: Zhu et al. (2021) - Project Manager's Emotional Intelligence
- High Expectations: Buschmeyer et al. (2022) - Expectation Management in AI
Plus 40 additional papers identified via Litmaps citation network analysis (N=44 total).
⚠️ Research Limitations
The study acknowledges specific boundary conditions:
1. Single-Sided Stakeholder Perspective
Focus on practitioners means client-side expectations remain theoretically inferred.
2. Digital Ecosystem Sampling Bias
Slack-based recruitment may underrepresent siloed sectors (defense, proprietary R&D).
3. Survivorship Bias
Experienced experts may have automated responses to friction that novices still struggle with.
🔒 Ethics & Anonymization
All participants provided informed consent before interviews. To preserve anonymity:
- Participants referenced as P1-P48 throughout all publications and materials
- All identifying details (company names, specific locations) removed
- Quotes used with permission and anonymized
- Data stored with enterprise-grade encryption
"The guys that work with us were not from data science... they do not know why the model was giving them that answer. We need this responsibility."
— P2, used with permission
🏛️ Affiliations
Rennes SB
Rennes School of Business - France
UpGrad
Engineering school - India
🤝 Collaborate With Us
Interested in applying the ATIS Framework in your organization or collaborating on research?
Youssef HARIRI • Rennes School of Business • youssef.hariri@rennes-sb.com