AI washing is a term to pass on basic rule-based system with no actual machine learning capabilities as AI. This is done to attract funding from venture capitalists (VCs). A startup can claim it is using AI for automation, though in reality it is not so. A chatbot can have no natural language processing (NLP) capabilities and may use manually programmed responses. These are exaggerated AI claims. They rebrand traditional software as AI-powered, without actual machine learning models or AI implementation. It is only pitching AI to create an illusion of advanced capabilities.

This issue prevails across various sectors — flow automation, data analytics which use simple IF-THEN rules or statistical methods. A marketing tool could claim to use AI, though in fact it uses predictive analytics applying pre-set formulae to detect trends, with no real-time learning.

Fintech startups claim they use AI powered bots to predict stock movements. Their systems may rely on basic statistical models or pre-defined rules. The rule could be buying when the prices fall by 10 per cent.

In the visual field, a startup could claim to use AI for image recognition or video enhancement. However, it may be using manual processes or traditional software. A video editing tool could be using pre-set filters rather than AI for automatic colour correction.

The scrutiny must be done by AI/ML experts. The experts could be drawn from elite institutions or by dedicated AI committees. The code base could be evaluated.