Discussions on importance of clean data, sustainable AI technologies
Experts at the QUEST-AI conference discussed the necessity of prioritizing clean and diverse data over sheer volume for reliable AI systems. They also explored the potential integration of quantum technology and nuclear-powered data centers to support future AI infrastructure.
Why it matters
As AI models become more pervasive, the quality of training data and the sustainability of the energy-intensive infrastructure required to run them are becoming critical global challenges.
Clean and diverse data are essential for reliable artificial intelligence (AI) systems, while synthetic data should be used within clearly defined limits, said AI technology expert Nilanjan Dey, Professor, Techno International New Town.
Speaking at the international conference on Quantum Enhanced Sustainable Technologies for AI Systems (QUEST-AI), organised by Andhra University Engineering College, Mr. Dey said machine-learning algorithms depended on data and that poor-quality data could produce flawed results. He said the focus should be on clean and diverse data rather than merely on “big data”, with greater attention also needed on “small data” where data availability was limited.
Mr. Dey said synthetic data could help when real-world data were scarce, but excessive dependence on it could increase errors in large language models (LLMs). He said its use should have clear limits.
Get smarter about the news
Sign up free for a feed built around what you actually care about, Dive Deeper research on any story, and the full text of every article.
Create free accountAlready have an account? Sign in