99% of UPSC aspirants are using AI wrong

Key idea: AI “assisted” learning vs AI “outsourced” learning

• The speaker argues that interaction with AI is not the same as learning.
• Don’t outsource thinking/cognition—only outsource friction, such as:
  • planning obstacles
  • concept explanation variety
  • feedback
  • quiz-style practice
• If you rely on AI to generate final answers/notes and never re-engage with the content, you’ll retain nothing.

---

Effective AI use for UPSC prep (core strategies)

1) Pick the right AI tool (any is fine)

• Gemini / Claude / GPT / Perplexity are all acceptable.
• The “best” tool is essentially the one you can access for free (or with minimal cost).

2) Use AI in the “right stages” of learning

A) Planning & Concepts (Stages 1–2)

• Make the plan once, then execute it.
• Use AI to:
  • explain concepts in multiple ways (ask for simpler explanations repeatedly)
  • reframe difficulty level (e.g., request explanation like a “smart 10th grader”; avoid childish/cartoon-level explanations)
• Don’t keep asking AI for new plans daily—execution beats perfect planning.

B) Reading & Note-making

• Reading must be done by you.
  • AI can summarize reports/books, but basic books should be reread.
• Make notes yourself the first time:
  • The benefit is the process of note-making, not just the content.
  • If AI writes your notes and you never mentally process them, information bypasses learning.
• After you’ve created notes, AI can help with “presentation”:
  • formatting, headings, making notes look good—especially for digital notes.

C) Interlinking subjects (Mind maps)

• Use AI to create mind maps that link a topic to relevant syllabus areas.
• Workflow:
  • Provide the topic
  • Ask AI to map it across syllabus areas to reveal connections you might miss

D) Current Affairs (don’t let AI “study” it directly)

• Avoid relying on AI for current affairs as-is because of:
  • cutoff/dated information
• Instead, use AI to integrate current affairs into static knowledge:
  • Ask: “How does this current affairs topic connect to my static subjects?”
  • Add it to notes using the linkage discovered (e.g., via mind maps)

E) Memorization & rote-heavy content

• Don’t ask AI for the answers directly—it won’t replace exam-ready memory.
• Use AI after memorization to:
  • quiz you and test recall
• The speaker’s emphasis: you must still sit and memorize/review repeatedly.

F) Answer writing (feedback, not auto-writing)

• Best use:
  • upload/post your drafted answer and ask AI to evaluate:
    • dimensions
    • frameworks
    • intro/conclusion improvements
    • PYQ-based dimension coverage
• Conceptually, the speaker suggests avoiding a raw “marks number” because marking subjectivity/normalization varies.
• Focus on demand fulfillment and missing dimensions.

G) Essay writing & revision

• Don’t outsource essay writing to AI.
• Use AI to evaluate:
  • whether your essay meets the demand of the question/prompt
  • whether you missed the core argument
• For revision:
  • use AI for question practice and thematic PYQ recall (the speaker mentions a “Prelims GPT” approach)
  • interrogate reasoning:
    • ensure the answer is correct for the right reasons, not by chance

H) Spoken English / Interview reflection

• Use AI voice mode to:
  • do mock speaking and get improvement pointers
  • get feedback on both content and spoken delivery
• The speaker recommends taking suggestions with a “pinch of salt”, but assumes many will be immediately useful.

---

Best practices: how to make AI actually improve retention

Use AI as a “hint generator” (Socratic learning)

• Worst way: asking AI to give the full answer.
• Better way:
  • ask for hints only
  • increase hints step-by-step until you figure it out
  • prefer questioning back (Socratic method) to force mental connections

Order matters

• First do the hard cognitive work yourself (read/understand/memorize/brainstorm).
• Then use AI to:
  • quiz you
  • provide feedback
  • help revise and strengthen gaps

Read vs work your brain

The speaker contrasts:

• Reading only = low retention
• Understanding/explaining/interrogating = much higher retention

AI is positioned as a tool for the “working brain” stages (quizzes, feedback, reasoning prompts).

---

Where not to use AI (explicit do-not-cases)

• Reading: you must read yourself (especially basic books)
• Thinking/decision-making: you should do it yourself
• Mock tests: must simulate exam conditions (OMR, no internet, strict time)
• Interview reflection: should be yours (AI for practice/feedback, not content replacement)
• Brainstorming: do it first; then ask AI to improve it
• Asking AI to write answers/essays directly: discouraged

---

Wellness / self-care angle (indirect but present)

• The message emphasizes mental ownership:
  • don’t become dependent on tools
  • maintain active cognition to avoid helplessness (e.g., the mindset: “If GPT couldn’t tell…”)
• Productivity outcome:
  • reduce wasted time/tokens by avoiding ineffective workflows (like making a PDF summary and never revisiting it)

---

Presenters / sources mentioned

Presenter/Speaker

• Not explicitly named in the subtitles (appears to be the YouTube creator/mentor speaking).

AI tools mentioned

• Gemini
• Claude
• GPT / ChatGPT
• Perplexity
• NotebookLM (called “Notebook Allen” in subtitles)

UPSC prep sources mentioned

• NITI Aayog documents
• Laxmikanth
• Spectrum
• NCERT (Geography)
• Geography PYQs from “Letts” (as referenced in subtitles)
• PYQs
• Indian Express and Hindu (used as example PDFs for topic extraction)