A research team led by Professor Enamul Hoque Prince of the School of Information Technology has won the Best Paper Award in the short category at IEEE Visualization 2025, an international conference on data visualization and analytics.
The award recognizes the paper The Perils of Chart Deception: How Misleading Visualizations Affect Vision-Language Models, co-authored by Prince and students Ridwan Mahbub, Mohammed Saidul Islam, Md Tahmid Rahman Laskar and Mizanur Rahman, with Mir Tafseer Nayeem as a collaborator.
The study investigates a timely question: if misleading charts can fool people, can they also fool today’s AI systems? The team examined how deceptive design choices – including truncated or inverted axes, stretched aspect ratios and unnecessary 3D effects – can distort a chart’s message. These techniques are known to mislead human viewers, sometimes subtly and sometimes dramatically. The study shows that popular AI models used to interpret visual data are often misled in the very same ways, producing incorrect explanations even when the underlying numbers remain unchanged.
Based on a large-scale evaluation of widely used AI tools, the research reveals that visual deception remains a challenge not only for human readers but also for automated systems increasingly used to support data analysis. The work highlights the need for more reliable and deception-aware AI tools as they become part of everyday decision-making.
Reflecting on the award, Prince stated, “As AI tools become more involved in helping people make sense of data, it is essential that they are not vulnerable to the same tricks that mislead humans. Our findings highlight an important gap that the research community must address. I am incredibly proud of my students for leading this work and grateful to our collaborators whose insights shaped this project.”
The paper was featured in the conference’s Best Paper session, marking a significant recognition for York University’s leadership in responsible AI research.
