Designing the AI-Era Data Analysis Textbook

Designing the AI-Era Data Analysis Textbook

An Interactive Learning Ecosystem for the AI-Enabled Workforce

Collaborative project with led by Prof Gábor Békés (CEU), co-author of Data Analysis for Business, Economics, and Policy and director of the Data Analysis and AI Lab, with Github, Inc and Cambridge University Press.

The Challenge

Data analysis education faces a fundamental mismatch: students are entering an AI-enabled workforce, but learning from static, print-first materials. While generative AI tools can explain code, clarify concepts, and generate examples, students lack structured guidance on how to use these tools effectively and responsibly in their learning.

Simply pointing students to ChatGPT or similar tools creates problems: results vary in quality, accuracy is inconsistent, and without proper context students may develop shallow understanding or even learn incorrect methods. We need to understand how different AI models perform across different educational tasks, and design materials that integrate AI as a complementary learning tool rather than a replacement for expertise.

What We’re Building

An open-source, AI-integrated textbook ecosystem that transforms how students learn applied data analysis. Building on the globally-adopted Békés-Kézdi Data Analysis for Business, Economics, and Policy textbook (used at 150+ universities in 40+ countries),

Check out a Pilot, details for Chapter 10: gabors-data-analysis.com/book/

we’re creating an interactive digital learning environment where AI supports four key activities:

  1. Explain: Intelligent tutoring that clarifies concepts and expands on material
  2. Examples: Context-relevant scenarios with real-world personas and applications
  3. Test: Adaptive assessments that respond to student understanding
  4. Code: Direct engagement with case studies via embedded GitHub Codespaces

The ecosystem combines textbook content, datasets, code, and AI tools in a unified platform that lowers access barriers while maintaining educational quality.

Our Research Approach

Understanding Current Practice

We’re surveying and interviewing instructors and students at community colleges, technical colleges, and liberal arts institutions worldwide to understand how AI is currently being used in data analysis education.

Benchmarking AI Performance

Different AI models excel at different tasks. We’re generating hundreds of AI outputs across our four application areas, then having human experts rate them on:

  • Correctness: Factual accuracy
  • Reliability: Consistency across time and geography
  • Faithfulness: Adherence to source material

This systematic evaluation will identify which models work best for which educational contexts, informing both our design and efficient use of AI in education more broadly.

This is the heart of the project: a collaboration of engineers and econometrics professionals testing AI’s role in education.

Pilot Testing

We’ll test prototypes in real classrooms, collecting feedback on learning outcomes and user experience to refine the approach iteratively.

Expected Outcomes

For Students: An accessible, interactive learning environment that builds both data analysis skills and AI fluency—preparing them for modern workplaces while ensuring they understand AI’s capabilities and limitations.

For Instructors: Evidence-based guidance on integrating AI into curricula, including model-specific performance benchmarks and cost considerations for different educational tasks.

For Institutions: A roadmap for modernizing data analysis education that balances innovation with pedagogical rigor, particularly valuable for community and technical colleges.

For the Field: Open-source tools, benchmark metrics, and documentation that other textbook authors and educators can adapt for their own materials.

Team & Partnerships

Principal Investigator: Gábor Békés (CEU), co-author of Data Analysis for Business, Economics, and Policy and director of the Data Analysis and AI Lab

Primary Partner: GitHub, Inc. (providing personnel and technical support)

Collaborators:

  • Eduardo de la Rubia (CEU, formerly senior Meta software engineer)
  • Michael Kozakowski (CEU Elkana Center of Learning)
  • Phil Good (Cambridge University Press)

The team combines expertise in economics, data analysis education, software engineering, learning sciences, and academic publishing.

Resources

Status & Next Steps

We’re currently seeking funding and partnerships to develop the prototype and conduct systematic benchmarking research.

Last updated: October 2025