Math AI is designed around real-world application, so the strongest Math AI explorations are grounded in genuine data — collected, scraped, or sourced from a credible dataset — rather than abstract theoretical modeling. If you’re a Math AI student, examiners want to see you interpreting numbers, not just manipulating symbols.
What Distinguishes a Strong Math AI Exploration
- Real, specific data (your own collected data is often stronger than a downloaded dataset, because it shows personal engagement)
- A clear statistical or modeling technique appropriate to that data
- Interpretation written in context — explaining what the numbers actually mean, not just stating them
- A thoughtful discussion of the limitations of your model or data collection method
15 Topic Ideas for Math AI (HL and SL)
Statistics & Data Analysis:
- Investigating whether there’s a correlation between screen time and sleep quality among students at your school (self-collected survey data)
- Analyzing whether home advantage is statistically significant in a sports league using real match data
- Comparing exam performance across different study techniques using a controlled survey and t-test analysis
- Investigating income inequality in a country using the Gini coefficient and Lorenz curve
Financial Mathematics: 5. Comparing the long-term value of different mortgage or loan repayment strategies using amortization models 6. Modeling the impact of compound interest frequency on long-term savings 7. Investigating whether a specific investment strategy (e.g., dollar-cost averaging) outperforms lump-sum investing using historical market data
Modeling & Optimization: 8. Using linear programming to optimize a real logistics problem (e.g., minimizing delivery costs for a small business) 9. Modeling traffic flow at a local intersection and testing an alternative light-timing scheme 10. Optimizing seating/space allocation in a venue using geometric and algebraic constraints
Probability: 11. Investigating the fairness of a lottery or game of chance using probability distributions and real outcome data 12. Modeling queue waiting times using probability distributions (Poisson or exponential models) with real data collected at a location
Networks & Graph Theory: 13. Using graph theory to find the most efficient route between locations (e.g., a delivery route or a public transport network) 14. Modeling a social network’s structure and analyzing connectivity patterns
Trigonometry & Geometry Applications: 15. Investigating the mathematics behind GPS positioning (triangulation) using real coordinate data
How to Collect Strong, Original Data
The highest-scoring Math AI explorations often involve primary data collection: running your own survey, timing your own experiment, or scraping publicly available data (sports statistics, government open data portals, financial market data) rather than relying on a textbook dataset that every other student in the world has access to.
Common Mistakes in Math AI Explorations
- Using a dataset that’s too small to justify meaningful statistical conclusions
- Applying a statistical test without checking whether its assumptions are actually met by the data
- Presenting results (a correlation coefficient, a regression equation) without interpreting what they mean in real-world terms
- Ignoring outliers or anomalies in the data instead of addressing them in the analysis
Final Tip
Before finalizing your topic, ask: «Can I realistically collect or find at least 20-30 data points for this?» Math AI explorations with too little data almost always struggle to justify a meaningful statistical conclusion, regardless of how interesting the topic is.