Video summary

10 AI SKILLS सीखो या बेरोजगार हो जाओ! | Ankit Avasthi Sir | Learn Free AI Tools | SKT Podcast

Main summary

Key takeaways

Technology

AI concepts & claims (friend vs threat)

  • AI is framed as a friend to humans, but risky if mishandled—often compared to a nuclear bomb scenario.
  • A future outlook is discussed (around 2047), suggesting humans may increasingly work under machines, with AI shaping employment and decision-making.
  • Despite automation, the emphasis remains that human creativity continues to matter.

Privacy, memory, and the “human dilemma”

The speaker contrasts how different systems handle user data:

  • Search engines (e.g., Google) are described as using personal history and logged-in behavior.
  • AI chat systems are described as remembering user context/data during a session, making them harder to fully delete/reset like browser history.

Risk framing includes:

  • Data retention
  • Identity inference
  • Potential for scams/blackmail, including misuse of stolen images or accounts

Job impact analysis (layoffs + sector disruption)

AI is presented as driving automation-driven disruption, especially in:

  • IT services/software roles (coding and development workflows)
  • Customer support / call centers

Counterpoint: the discussion also highlights that AI creates new opportunities for people who learn to use tools and shift roles, such as:

  • building services
  • creating content pipelines
  • managing operations with AI assistance

“Prompt engineering” and tooling workflow: Context–Ask–Rule–Example (CARE)

A tutorial-style prompting framework is presented:

  • Context → who/what you are
  • Ask → what you want
  • Rule → constraints/style/safety boundaries
  • Example → a sample output to emulate

Claim: better prompts produce higher-quality results more reliably.

The transcript also mentions building multiple “GPTs”/custom agents for different tasks (details are noisy).

Automation examples / product-like features mentioned

  • ChatGPT / OpenAI: positioned as a “leader,” widely accessible, with abilities to answer, write, and integrate tasks via tools/plugins.
  • Sora: mentioned as a media-generation AI product.
  • Gemini / Perplexity / Claude / Grok: compared by use-case:
    • ChatGPT: general assistance + ecosystem
    • Perplexity: described as “real-time” search-style responses
    • Claude: described as strongest for using your own files/computers (e.g., “cowork” and document/file workflows)
    • Grok: described as fast image/video generation with a different writing style
  • NotebookLM (Google tool): described as a workflow to:
    • paste a YouTube link
    • convert content into notes, points, bullets, and study slides
  • Claude + file/C-drive operations: described as enabling AI-driven automation of folders/files and spreadsheet creation.

“Website in 15 minutes” demo (practical tutorial claim)

The host claims to create a website quickly (a domain like ankitavasthi.com is mentioned). The claim is that AI can:

  • gather public internet information
  • assemble:
    • website content
    • clickable links
    • structured pages

It’s also discussed that tasks once requiring developers/coding/WordPress themes can be accelerated by AI.

“Vibe coding” / coding automation (transforming development workflows)

Coding is described as shifting from manual work to AI-driven generation and modification:

  • AI converts prompts into code-like outputs
  • developers can request changes to existing code via new instructions

A simplified analogy is used to explain why models can process large knowledge, referencing concepts like GPU/transformers.

Cutoff dates & credibility / “knowledge not updated” risk

Models are explained as having a knowledge cutoff date (example mentioned: Aug 2025).

Implications:

  • older facts can be wrong or outdated
  • the transcript uses informal examples (especially around politics) to make the reliability point

Perplexity is praised as mitigating this through search-based real-time retrieval.

Sector-specific guidance (where AI is most useful)

  • Students / exam prep (UPSC focus)
    • AI can act like a counselor/mentor, generating study drafts and guidance
    • example prompting emphasizes context + persona + time constraints
    • NotebookLM is presented as especially effective for turning long videos into study artifacts
  • Content creators
    • AI suggestions for thumbnails/titles/design
    • generating community posts after analyzing video content
  • Email support / productivity
    • AI described as helping by pre-drafting replies for Gmail/work emails
  • Call centers / service industry
    • AI call agents described as handling many calls quickly and understanding multiple languages
  • Local businesses / marketplaces
    • encourages small vendors to digitize (web presence/group selling) and compete via tech-enabled distribution

AI tool list (short recommendations mentioned)

  • Luma / Lardo / Suno / Sora: music/video/image generation (names are noisy in the transcript)
  • Grok: image/video generation; fast outputs
  • DeepSeek: described as strong for research
  • NotebookLM: video → notes/study points
  • Notion / automation: task management + AI workflows
  • Microsoft Copilot: described as assisting computer/cloud workflows

Safety/ethics points raised

  • AI increases cybercrime risk (strongly stated).
  • Caution: don’t trust AI outputs for high-stakes medical/critical domains.
  • Mentions AI being used in warfare scenarios (drones + automation) and related geopolitical impacts.

Main speakers / sources

  • Ankit Avasthi Sir: exam educator (also referenced for an AI tools learning/course perspective)
  • Shalini Kapoor Tiwari (SKT): podcast host (“SKT means I Shalini Kapoor Tiwari”)

Original video