Archive · 3 Sep 2026
Experts Urge Clean, Diverse Data for More Reliable AI
AI systems learn from data, much like students learn from examples.
Professor Nilanjan Dey said that having more data is not always enough.
The data must also be accurate, clean, and varied.
If an AI system sees only repeated examples, it may struggle with new situations.
Data from many different real-world conditions can help AI work better.
Dey said AI could help healthcare workers manage their workload.
He noted reports estimating that doctors in India spend an average of 2.8 minutes with each patient, compared with 48 seconds in Bangladesh.
He also said synthetic data may influence the next stage of AI development.
Professor Nilanjan Dey warned that unreliable or unscientifically collected data can produce flawed AI results.
He said increasing data volume alone does not necessarily improve artificial intelligence systems.
Diverse data covering more variables and real-world situations can make models more robust.
Limited or repetitive training datasets may cause models to perform poorly in unfamiliar situations.
Dey highlighted healthcare applications, synthetic data, and technology that could reduce pressure on doctors.
- Who
- Professor Nilanjan Dey of Techno International New Town.
- What
- He argued that reliable, clean, and diverse data is more important than simply collecting larger amounts of data for AI systems.
- When
- Wednesday, during the ongoing international conference QUIST-A.
- Where
- AU Engineering College in Visakhapatnam.
- Why
- To explain how data quality and diversity affect AI reliability, including potential healthcare applications.
- Speaker
- Professor Nilanjan Dey
- Conference
- International conference Quantum Enhanced Sustainable Technologies for AI Systems (QUIST-A)
- Venue
- AU Engineering College, Visakhapatnam
- Core message
- More data does not necessarily mean better AI.
- Recommended focus
- Clean data and variable data rather than only big data
- Healthcare statistic
- A recent report cited average doctor-patient times of 2.8 minutes in India and 48 seconds in Bangladesh
- Emerging development
- Synthetic data could shape the next phase of AI