Learn with Anil · UPSC Civil Services PYQsMains PYQs →Subscribe ₹149
INS Vikrant and INS Vikramaditya sailing with their carrier battle group

Photo: INS Vikrant and INS Vikramaditya, Indian Navy, GODL-India

UPSC Civil Services (Main) 2026 · GS Paper III

UPSC Mains 2026 GS 3 · Q16: model answer

General Studies Paper III: Technology, Economic Development, Bio-diversity, Environment, Security and Disaster Management · Science and Technology.

Q1615 marks250 words2026 · GS Paper III

What is agentic Artificial Intelligence (AI)? Explain its working. Describe its applications with suitable examples. Discuss the advantages, risks and challenges associated with agentic AI systems.

Syllabus: Science and Technology · IT, computers and artificial intelligence

Model answer · free

Agentic AIAutonomy and tool useLarge language modelsResponsible AI governance

Agentic Artificial Intelligence refers to AI systems that pursue a goal with limited human supervision: they plan the steps, use tools such as search, software and payment systems, act in digital or physical environments and adjust to the results. A chatbot answers a prompt; an AI agent completes a task, such as booking a trip end to end. Built mostly on large language models (LLMs), it is the next stage after generative AI.

1. How agentic AI works

  • Goal and context: a user sets an objective; the agent gathers information from documents, the web or sensors.
  • Reasoning and planning: the LLM breaks the goal into sub-tasks and decides the next action.
  • Tool use and action: it calls software interfaces (APIs), runs code, fills forms, sends messages or controls devices and robots.
  • Memory: it keeps track of steps done and remembers preferences across sessions.
  • Feedback loop: it checks results, corrects errors and repeats until the goal is met, asking a human to approve key actions; in multi-agent systems specialised agents divide the work.
User sets a goalAgent plans thestepsAgent acts throughtoolsAgent observesresults andupdates memoryAgent correctscourse or seeksapprovalThe agent loop
AspectGenerative AI (chatbot)Agentic AI
OutputText, image or code on requestCompleted tasks and actions
AutonomyResponds to each promptPlans and acts over many steps
ToolsFew or noneBrowses, runs code, uses software and devices
Human roleGuides every stepSets the goal, approves key actions
Generative AI vs agentic AI

2. Applications with examples

  • Software and IT: coding agents write, test and fix code while developers review the changes; IT agents monitor systems and fix routine faults.
  • Citizen services: agents can handle grievances, check eligibility and fill forms in Indian languages, building on Bhashini and digital public infrastructure such as DigiLocker and UPI.
  • Business: agents resolve customer queries, reconcile invoices and flag fraud.
  • Health and science: agents schedule patients, draft clinical notes and run literature reviews for drug discovery.
  • Agriculture: agents can combine weather, soil and market data to advise farmers.

3. Advantages

  • Productivity: routine multi-step work is done faster and round the clock, freeing people for judgement-based tasks.
  • Inclusion: voice agents in local languages can help citizens with low literacy and persons with disabilities.
  • Scale for small firms: MSMEs can use agents for accounts, marketing and compliance that they could not staff.

4. Risks and challenges

  • Compounding errors: a "hallucination" in a chatbot is a wrong answer; in an agent it can become a wrong payment or a deleted file.
  • Security: prompt injection (hidden instructions in web pages or emails that hijack an agent), misuse for cyber attacks and scams.
  • Accountability: unclear liability among developer, deployer and user when an agent errs.
  • Privacy: agents need wide access to emails, files and accounts, which must comply with the DPDP Act, 2023 and its 2025 Rules.
  • Jobs: routine white-collar work in IT services and back offices may shrink.
  • Concentration: dependence on a few foreign model providers and energy-hungry data centres.

India's AI Governance Guidelines (November 2025) favour "trust is the foundation" and "people first" without a separate AI law, backed by the IndiaAI Mission (₹10,371.92 crore) and an AI Safety Institute. Human approval for high-stakes actions, permission limits, audit trails and skilling can make agentic AI a safe force multiplier for Viksit Bharat.

Every model answer here is free

The ₹149 All-in-One Pass opens the rest of the site for 12 months: UPSC Prelims GS Paper I and CSAT with explanations, practice quizzes and timed tests, UPSC CAPF and EPFO, the previous papers of every APPSC, TSPSC, police and SSC exam, every mock test series and the study books.

₹149 · valid for 12 months from the day you pay, on every exam · pay by UPI, card or net banking

Get it for ₹149