Wednesday, 2 September 2026

Generative AI & Large Language

 

Generative AI & Large Language Models

 Revision Notes

Source: IBM Skills build — Foundations in Generative AI, Introduction to LLMs, AI Ethics & Workplace AI

https://cmis5.anudip.org/auth/ibm-student-nomination

 1. Generative AI Fundamentals

    Foundation Models — Large-scale models trained on massive (terabyte-scale) unstructured data using self-supervised learning (predicting patterns without human labels).

    Large Language Models (LLMs) — A type of foundation model specializing in generating and analyzing human language.

    Core Mechanism — LLMs work by predicting the next word in a sequence, generating responses one word at a time.

    Tuning — Adapting a foundation model to a specific task by introducing a small amount of labeled data (e.g., classification).

    Prompting — Applying the model to new tasks without additional training, using instructions or questions (e.g., asking if a sentence is positive or negative).

    Advantages — High performance (vast training data) and productivity gains (requires less labeled data).

    Disadvantages — High compute costs to train/run; trustworthiness issues (bias, toxic content, lack of transparency on training datasets).

    Domains — Language, Vision (images), Code, Chemistry, and Climate.

    Generative AI Tasks — Idea generation, creating content drafts at scale, problem solving, summarizing/categorizing data, context understanding, and prediction.

 2. Prompt Engineering Techniques

    Be Specific

    Vague prompts ("Summarize this") yield vague outputs. Add explicit details about length, focus, and tone.

    Include Examples (Few-shot Prompting)

    Providing examples of correct classifications (e.g., positive vs. negative reviews) reduces misclassification.

    Refinement Techniques

    Add the format the output should take (e.g., "list of bullets").

    Add the role the AI should take and the expertise to use.

    Add information describing exactly what you want the AI to do.

    Prompt Components

    Task • Context • Role • Constraints • Output format.

 3. AI in the Workplace

    Automation

    Handling repetitive, predictable tasks (e.g., sorting files, generating routine reports).

    Augmentation

    AI supporting human thinking, decisions, and creativity (e.g., smart assistants helping prioritize meetings).

    Human Judgment is the Ultimate Authority

    AI provides speed; humans provide direction, context, and accountability. AI output must always be reviewed and approved by a human — never treated as final.

    Case Study — IBM Bob (AI Developer Partner)

    Bob builds code foundations, suggests fixes, runs safety checks, documents code, and refines old code.

    The developer always investigates, polishes, and makes the final decisions.

    Other Roles

    Project Managers: AI organizes transcripts; managers prioritize and message stakeholders.

    Cybersecurity: AI prioritizes alerts; analysts evaluate true risk.

    Data Analysts: AI translates natural language to queries; analysts validate assumptions.

    Educators: AI generates content; educators shape meaning and build relationships.

 4. Ethical Considerations for Generative AI

    Three Pillars of AI Ethics

    Transparency: Users understand how AI works and can trust its outputs.

    Accountability: People and organizations take responsibility for AI outcomes.

    Fairness: Preventing biases from distorting results.

    Common Ethical Risks

    Plagiarism — AI-generated content closely resembling an artist's/writer's work.

    Bias / Lack of Fairness — training on historical patterns lacking diversity (e.g., favoring degrees or past-hire similarity).

    Data Privacy Violations — using real, anonymous customer data in training inputs.

    Harmful Misinformation — incomplete/outdated data used for decisions such as healthcare diagnoses.

    Lack of Transparency — undisclosed algorithms filtering content, leading to distrust.

    Strategies

    Train on diverse demographic data, regularly test for equitable outcomes, provide transparency, and manage bias.

 5. IBM Granite Models & Data Analytics

    Granite Code

    Best for generating efficient code, automating routine tasks, and explaining complex code for onboarding.

    Granite Instruct

    Best for analyzing customer reviews to identify trends and provide summarized insights.

    Granite Multilingual / Japanese

    Specialized for multi-language and Japanese text tasks (not code).

    Data Analytics Transformation

    Gen AI shifts analytics from reporting on the past to collaborating in real time.

    Users query in natural language (e.g., "Which creators grew fastest?").

    Generates summaries, charts, code scripts, narratives, and dashboards — empowering non-coders and data scientists alike.

    Use Case — Personalization

    E-commerce companies use LLMs to recommend products based on individual customer preferences and purchase history.

 6. Quick-Reference Answer Key

Scenario Cue

Correct Answer

Personalization based on purchase history

Personalization use case

Generate efficient code + explain code for onboarding

Granite Code

Analyze reviews for trends & insights

Granite Instruct

Model misclassifies without examples given

Include examples

Vague summary from vague prompt

Be specific

Naming/generating catchy ideas

Idea generation

Producing response word-by-word

Prediction step

Determining intent/tone of a request

Contextual understanding

100 product descriptions quickly

Creating content drafts at scale

Summarizing/categorizing thousands of entries

Problem solving

Wants output as bullet list

Add output format

Wants more audience-relevant detail

Add role & expertise

Response too broad

Add exact task detail

Diverse training + bias testing

Fairness

Tool favors certain universities despite balanced data

Managing bias

Favors candidates similar to past (non-diverse) hires

Lack of fairness

Undisclosed content-filtering algorithm

Providing transparency

Chatbot leaks another customer's data

Data privacy violation

 https://cmis5.anudip.org/auth/ibm-student-nomination

— End of Notes —


No comments:

Post a Comment

NPTL Ethics Notes

Ethics in Engineering Practice (Week 3, Lectures 11–15) course material. It synthesizes abstract moral frameworks with tangible IP mechani...