UPSC Mains 2026 GS3 Q16: model answer

Agentic AI pursues a goal rather than just answering: it plans steps, uses tools, acts and checks results with limited human help. A chatbot gives train timings; an agent books, pays and updates your calendar.

How it works: goal set → a large language model plans steps → it calls tools (search, databases, payment APIs, code) → observes results, stores memory, revises → loops until done, with human approval at checkpoints.

Applications

• Coding agents that write, test and fix code.
• Banking and customer service resolving complaints end to end.
• Multilingual assistants guiding farmers through scheme forms, building on Bhashini.
• Health triage and follow-up; literature and lab simulation in science.

Advantages: productivity, round-the-clock service, Indian-language reach, help for small firms.

Risks

• Accountability for wrong payments or decisions.
• Security: prompt injection, tool misuse.
• Errors compounding over many steps.
• Privacy under the DPDP Act, 2023.
• Jobs in IT services and BPO.
• Compute costs and dependence on foreign models.

The IndiaAI Mission's shared GPUs and domestic models, plus clear rules on oversight and liability, can let India use agents safely rather than only import them.