Artificial intelligence agents are being designed to do more than answer questions. They can potentially organise travel, manage bookings, draft emails and complete everyday tasks with limited human intervention. But there is one major obstacle to their wider adoption: whether people are comfortable letting AI act on their behalf. In a recent post on X, Manu Kumar Jain, former Xiaomi India head and CEO of G42 India, argued that the most capable AI agent will not necessarily emerge as the winner. Instead, success will depend on which agent users trust enough to give greater control over their digital lives.
Why Trust Could Matter More Than AI Benchmarks
Jain said he had been using four to five leading AI agents but had not given any of them access to his primary email account, credit card or UPI. His concern was straightforward: what happens if an agent sends the wrong email or makes an incorrect payment?
This hesitation highlights a gap between what AI agents promise and what users are willing to let them do. An assistant might be capable of planning a trip or organising bookings, but without the necessary permissions, users still have to supervise each step and complete parts of the task themselves. Jain compared such an agent to a capable intern who has not yet earned the user's trust. He argued that people generally delegate responsibilities gradually, giving someone more authority as they demonstrate reliability. AI agents, he suggested, will need to earn trust in much the same way.
ALSO READ: ChatGPT Gets Locked Chats To Keep Private Conversations Away From Prying Eyes
How AI Agents Can Earn Users' Trust
According to Jain, AI agents should seek approval before taking important actions, offer ways to undo mistakes and maintain complete audit trails and financial records. These safeguards could help users understand what an agent has done, why it acted and how to reverse an unwanted action. The concerns are particularly relevant when agents handle sensitive information or money. India's cybersecurity agency CERT In warned in an October 15 advisory that some AI agents had exceeded their assigned permissions and accessed sensitive information in live systems.
India's AI Payments Push Raises The Stakes
The question of trust is also becoming important for digital payments. Reuters reported in September that the National Payments Corporation of India was developing a registry to verify and monitor AI agents conducting UPI transactions under a proposed framework. Such systems could help establish which agents are authorised to make payments and improve transaction monitoring. However, questions around security, accountability and liability still need to be addressed.
The next phase of AI competition may not be decided by which model scores highest on a benchmark, but by which agent people are comfortable allowing into their inboxes, calendars and financial lives. For AI companies, earning that confidence could prove just as important as making their technology smarter.
ALSO READ: Deepinder Goyal Teases 'Dark.ai' From Eternal Ahead Of Q2 Results
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.
