AI and Business Environment: formula sheet

Key formulas

Relationship of AI, ML and DL

AI ⊃ ML ⊃ DL

Deep learning is a subset of machine learning, which is a subset of AI.

Types of AI by capability

Narrow AI → General AI → Super AI

Only narrow AI exists today. General and super AI are not yet achieved.

AI vs traditional software

Traditional: rules + data → output; AI/ML: data + expected output → rules (model)

Traditional software is programmed. AI learns patterns from data.

Core ethical concerns

Bias + Privacy + Opacity + Accountability + Misuse

Use this as a checklist to open any answer on ethical issues of AI.

Responsible AI principles

Fairness, Transparency, Accountability, Privacy, Safety, Human oversight

Wording varies by source. Learn the ideas and give a one-line meaning for each.

India's legal position

No single AI law; existing laws + DPDP Act, 2023 + policy guidance

Do not say India has a dedicated AI Act. Say AI is covered by existing laws and guidance.

Global approach

EU AI Act = risk-based classification of AI systems

Higher risk means stricter obligations. Voluntary principles from OECD and UNESCO are not binding.

NITI Aayog strategy

National Strategy for AI (2018) = #AIforAll

Vision: inclusive AI for social good. Five focus sectors: healthcare, agriculture, education, smart cities and infrastructure, smart mobility and transportation.

IndiaAI Mission

Approved March 2024 | outlay over ₹10,000 crore | nodal ministry: MeitY

Pillars: computing capacity, foundation models and datasets, startup support, skilling, safe and trusted AI.

Answer frame

Scheme → Year → Body → Aim → Focus areas → Challenge

Use this order for any written question on AI policy.

Quick revision

• Artificial intelligence means machines or software performing tasks that normally need human intelligence, such as learning, reasoning and understanding language.
• Machine learning lets systems learn patterns from data instead of following only fixed rules.
• Natural language processing helps machines understand and produce human language, as in chatbots.
• Computer vision lets systems interpret images and video.
• Business uses: customer service, marketing personalisation, fraud detection, demand forecasting, hiring support and process automation.
• Benefits: speed, lower cost over time, accuracy, round-the-clock service and better data-based decisions.
• Challenges: data quality, cost, skill gaps, job disruption, security risks and wrong or unexplained outputs.
• Ethical concerns: bias, privacy, lack of transparency and unclear accountability.
• Regulation tries to protect people while still allowing innovation.
• For India, learn the policy points from your study material and link them to growth, skills and responsible use.
• In every answer, define first, explain with an example, then conclude.

Common mistakes

• Saying ChatGPT-type tools or voice assistants are general AI. Fix: Remember that all AI in use today is narrow AI, however impressive. General AI is still a goal.
• Using AI, ML and DL as the same thing. Fix: Write the nesting AI ⊃ ML ⊃ DL and define each in one line.
• Saying machine learning and NLP are the same thing. Fix: Remember ML is a method of learning from data and NLP is a field about human language. NLP often uses ML, but ML also works on numbers and images.
• Calling every chatbot a generative AI tool. Fix: Say that simple chatbots follow fixed rules. Only some use generative AI to create replies.
• Listing tools instead of applications, such as only writing machine learning or chatbots. Fix: Always tie the tool to a business task. Write chatbots for customer queries, not just chatbots.
• Giving the same vague point, such as saves time, for every function. Fix: Learn one or two specific tasks per function, such as résumé screening for HR and fraud detection for banking.
• Writing only benefits or only challenges. Fix: Always cover both sides unless the question asks for one, and add a conclusion.
• Saying AI will end all jobs. Fix: Say AI displaces routine tasks but also creates new roles, so reskilling is needed.
• Saying India has a dedicated AI Act. Fix: Write that India relies on existing laws, the DPDP Act, 2023 and policy guidance, not a single AI statute.
• Treating the DPDP Act as a law only about AI. Fix: State that it is a general personal data law that affects AI because AI uses personal data.

Exam tips

• Definition plus types is the most likely pattern. Prepare a ready 5-line definition and a 3-type list.
• For differentiate questions, give at least four points in a clear two-column layout.
• Always add an Indian business example. It separates your answer from a plain memorised one.
• Do not name any existing product as general or super AI.
• Spend the 15 minutes of reading time marking which questions on this chapter you can answer fully.
• Keep a one-line definition and one business example ready for each of the six technologies.
• For distinction questions, write two or three comparison points, not just two definitions.
• Link each technology to a business function, as the chapter covers applications and challenges as well.