AICT Lecture 03 -Command Prompts in AI.pptx

Application of Information and
Communication Technology
Lecture - 03
Introduction to Artificial Intelligence and
Prompt Generation

Introduction to Artificial
Intelligence (AI)
• Definition of AI: Artificial intelligence (AI) is a branch of computer science
that deals with the creation of intelligent agents, which are systems that can
reason, learn, and act autonomously.
• The Birth of Artificial Intelligence:
• 1950s: The term "Artificial Intelligence" coined by computer scientist John
McCarthy. Alan Turing publishes his seminal paper "Computing Machinery
and Intelligence," which introduces the concept of the Turing test as a
measure of machine intelligence.
• 1956: Dartmouth Conference marks the official birth of AI as a field of study.
• 1950s-1960s: Early AI pioneers explore symbolic reasoning and problem-
solving. The first AI programs are developed, including ELIZA, a chatbot that
can simulate conversation, and General Problem Solver, a program that can
solve a wide range of problems.
• 1965-1974: Development of expert systems and rule-based AI.

Introduction to Artificial
Intelligence (AI)
• 1997: IBM's Deep Blue defeats world chess champion Garry
Kasparov.
• 2000s-Present: AI renaissance fueled by big data, improved
algorithms, and increased computing power.
• 2010s: Breakthroughs in machine learning.
• 2020s: Continued advancements in deep learning, reinforcement
learning, and ethical AI considerations.
• Types of Artificial Intelligence:
• Narrow AI (Weak AI): Specialized in a specific task
• General AI (Strong AI): Possesses human-like cognitive abilities.
• Superintelligent AI: Surpassing human intelligence (theoretical).

AI As Job Killer or Job Creator!
• Navigating the Impact on Employment: AI is often portrayed as a job
killer, but it is more likely to be a job creator. AI will create new jobs in
areas such as AI development, maintenance, and support. AI will also
automate many tasks that are currently performed by humans, which will
free up humans to focus on more creative and strategic work.
• Emerging Job Roles in AI: Chatbots and Virtual Assistants, Predictive
Customer Support, Personalized Recommendations, Medical Imaging
Diagnostics, Drug Discovery, Personalized Medicine also transforming
finance with Intelligent Systems (i.e. Fraud Detection & Credit Scoring
etc.) and resolving legal issues etc.
• Ethical Considerations and Responsibility: Guiding AI with Ethics,
Bias and Fairness, Privacy Concerns and Transparency and
Accountability evaluation of policies and procedure etc.

Application of AI in today’s
world:
Here are some examples of AI in use today:
• Self-driving cars: AI is used to power the self-driving features in cars such as lane
keeping assist and adaptive cruise control.
• Virtual assistants: AI is used to power virtual assistants such as Siri, Alexa, and
Google Assistant. These assistants can answer questions, control smart home devices,
play music, and more.
• Medical diagnosis: AI is used to develop systems that can help doctors diagnose
diseases and recommend treatments.
• Fraud detection: AI is used to develop systems that can detect fraudulent transactions
and other types of financial fraud.
• Product recommendation systems: AI is used to develop systems that can
recommend products to customers based on their past purchases and browsing
behavior.
• Image recognition: AI is used to develop systems that can identify and classify objects
in images. These systems are used in a variety of applications, such as facial
recognition, self-driving cars, and medical imaging.

Application of AI in today’s
world:
• Speech recognition: AI is used to develop systems that can transcribe
speech to text. These systems are used in a variety of applications, such as
voice assistants, dictation software, and closed captioning.
• Social media: AI is used to recommend content to users, detect spam and
abuse, and personalize the user experience.
• Customer service: AI is used to develop chatbots that can answer customer
questions and resolve issues.
• Education: AI is used to develop personalized learning programs and to
assess student progress.
• Robotics: AI is used to develop robots that can perform tasks autonomously.
These robots are used in a variety of industries, such as manufacturing,
logistics, and healthcare.
• Gaming: AI is used to develop video games with more realistic and
challenging opponents. AI is also used to generate realistic graphics and
animations.
• Finance: AI is used to develop systems that can make investment decisions,
detect fraud, and automate risk management tasks.

Advanced AI Tools:
• OpenAI's GPT-3.5: Chat Generative Pre-trained Transformer,
is a large language model-based chatbot.
• Google Cloud AI Platform: Google Cloud AI Platform is a
suite of cloud-base.
• Microsoft Azure Machine Learning: Microsoft Azure
Machine Learning is a suite of cloud-based AI services.
• Amazon Web Services (AWS) Machine Learning: Amazon
Web Services (AWS) Machine Learning is a suite of cloud-
based AI services.

Prompt Engineering
Prompt engineering as a strategic approach to interacting with AI,
especially large language models like ChatGPT, to produce precise,
useful, and contextually relevant answers.
To explore how different prompt engineering techniques can be
used to achieve different goals. ChatGPT is a state-of-the-art
language model that is capable of generating human-like text.
However, it is vital to understand the right way to ask ChatGPT in
order to get the high-quality outputs we desire.

Introduction to Prompt
Engineering Techniques
• Definition: Prompt engineering involves formulating well-
structured inputs that enable AI to generate text tailored to the
user’s needs, focusing on three core components – task,
instructions, and role – to ensure relevancy and quality in AI
output.
• Purpose: Controlling AI outputs to suit specific needs and tasks.
• Components: Task, instructions, and role as foundational
elements in crafting prompts.

Instructions Prompt Technique
• Description: Instruction prompts are carefully crafted inputs that
include clear, specific instructions that guide the AI’s response
structure and tone, helping ensure the generated content is
relevant and aligned with the user's purpose.
• Example Prompt Formula: “Generate [task] following these
instructions: [instructions].”
• Examples: Generating customer service responses with
instructions to keep responses professional and accurate.

Role Prompting
• Explanation: Role prompting assigns the AI a specific role or
persona, such as a teacher or customer service agent, to
influence the tone and content of the response, enhancing its
suitability for particular audiences or contexts.
• Example Prompt Formula: “Generate [task] as a [role].”
• Use Case: Generating a product description with the AI in the role
of a marketing representative.

Types of Prompts For (ChatGPT)
• Standard Prompts
• Zero Shot, One Shot & Few Shot Prompts
• “Let’s Think About This” Prompt
• Self-Consistency Prompt
• Seed-Word Prompt
• Knowledge Generation Prompt
• Knowledge Integration Prompt
• Multiple Choice Prompts
• Interprétable Soft Prompts
• Controlled Generation Prompts
• Question-Answering Prompts
• Summarization Prompts
• Dialogue Prompts
• Adversarial Prompts
• Clustering Prompts
• Reinforcement Learning Prompts
• Curriculum Learning Prompts
• Sentiment Analysis Prompts
• Named Entity Recognition Prompts
• Text Classification Prompts
• Text Generation Prompts

Standard Prompts
• Definition: Standard prompts are straightforward commands that
specify a single task without additional role or context, allowing for
a baseline response that focuses on clarity and conciseness.
• Example Prompt Formula: “Generate a [task].”
• Example: Summarizing a news article or writing a review of a
smartphone.

Zero, One, and Few-Shot
Prompting
• Overview: These techniques refer to the number of examples
provided in the prompt. Zero-shot offers no examples, one-shot
offers one example, and few-shot offers a limited number, guiding
the AI on response style or content expectations.
• Example Prompt Formula: “Generate text based on [number]
examples.”
• Examples: Comparison of product descriptions with different
levels of example support.

“Let’s Think About This” Prompt
• Purpose: This technique encourages reflective AI responses by
prompting a deeper exploration of a topic, useful in generating
thought-provoking content, personal reflections, or discussions.
• Example Prompt Formula: “Let’s think about this: [topic].”
• Examples: Writing a reflective essay on personal growth or
discussing climate change impacts.

Self-Consistency Prompt
• Technique: This approach ensures that generated responses are
consistent across similar inputs, avoiding contradictions and
maintaining logical coherence in outputs, ideal for tasks that
require reliable factual or informational consistency.
• Example Prompt Formula: “Ensure the following text is self-
consistent: [input text].”
• Example: Checking factual consistency in product information
within a review.

Seed-Word Prompt
• Description: The seed-word technique focuses the AI on a
specific word or phrase, subtly guiding the content and theme of
the response without rigid instructions, often used to keep
responses aligned with a core topic.
• Example Prompt Formula: “Please generate text based on the
following seed-word: [word].”
• Examples: Using the seed-word “innovation” in a product
description.

Knowledge Generation Prompt
• Purpose: This technique prompts the AI to generate new
information or insights based on its training, allowing it to explore
topics without relying on examples, especially valuable in
knowledge-based applications.
• Example Prompt Formula: “Generate new information about
[topic].”
• Example: Eliciting information on renewable energy
advancements.

Knowledge Integration Prompt
• Usage: Knowledge integration prompts merge existing AI
knowledge with newly introduced information, creating a more
holistic response by incorporating multiple sources or aspects of
the topic.
• Example Prompt Formula: “Integrate the following information
with existing knowledge about [topic]: [new information].”
• Example: Integrating smartphone features with market trends.

Multiple Choice Prompts
• Technique: Multiple-choice prompts restrict the AI’s responses to
a set of predefined options, helping streamline responses to
known choices, useful in surveys, sentiment analysis, and
categorization tasks.
• Example Prompt Formula: “Answer the question by selecting
one of these options: [options].”
• Use Cases: Multiple choice questions, sentiment analysis.

Interprétable Soft Prompts
• Description: Interpretable soft prompts strike a balance between
controlled input and AI flexibility, ensuring the AI adheres to
certain themes while allowing for natural variation, commonly
applied in creative storytelling.
• Example Prompt Formula: “Generate text based on these
characters [characters] and theme [theme].”
• Example: Crafting a story with characters following a specific
theme.

Controlled Generation Prompts
• Explanation: Controlled generation prompts limit the AI to specific
vocabulary, style, or structure constraints, ideal for formalized
tasks like legal, academic, or technical writing where consistency
is paramount.
• Example Prompt Formula: “Generate text based on the following
template: [template].”
• Example: Structured responses in customer service interactions.

Question-Answering Prompts
• Technique: Question-answering prompts are designed to elicit
factual or informative responses, guiding AI to retrieve or infer
relevant information for clear, concise answers.
• Example Prompt Formula: “Answer the following question:
[question].”
• Use Case: Generating accurate responses to complex questions.

Summarization Prompts
• Purpose: Summarization prompts help condense large volumes
of text into concise summaries that retain key details, making it
ideal for compressing information in meetings, articles, or reports.
• Example Prompt Formula: “Summarize the following text: [text].”
• Example: Summarizing meeting notes or lengthy articles.

Dialogue Prompts
• Description: Dialogue prompts simulate conversations by
defining characters, roles, and context, guiding AI to produce
natural interactions, useful in applications like customer support
and story creation.
• Example Prompt Formula: “Generate dialogue between
[characters] in the context [context].”
• Example: Crafting customer support interactions.

Adversarial Prompts
• Purpose: Adversarial prompts test AI response robustness by
introducing ambiguity or complexity, valuable in assessing AI
performance and reliability under challenging conditions.
• Example Prompt Formula: “Generate text that is challenging to
classify as [label].”
• Example: Creating challenging prompts in sentiment analysis.

Clustering Prompts
• Usage: Clustering prompts direct the AI to group similar items
based on specific attributes, allowing for categorization by
sentiment, topics, or themes, valuable in data analysis and
content organization.
• Example Prompt Formula: “Cluster the following data based on
[criteria].”
• Example: Grouping reviews based on sentiment or topics.

Reinforcement Learning Prompts
• Technique: Reinforcement learning prompts help AI improve
responses based on feedback, enabling continuous learning and
adaptation, essential for refining tone, style, or accuracy over
time.
• Example Prompt Formula: “Use reinforcement learning to
generate consistent responses in [style].”
• Example: Improving conversational tone over time.

Curriculum Learning Prompts
• Concept: Curriculum learning prompts start with simple tasks that
build in complexity, guiding AI through a structured learning path
to master sophisticated tasks gradually, applied in education or
skill training.
• Example Prompt Formula: “Generate responses using
curriculum learning in the order: [tasks].”
• Use Case: Gradual skill-building in complex tasks.

Sentiment Analysis Prompts
• Purpose: Sentiment analysis prompts help AI assess emotional
tone within text, classifying it as positive, negative, or neutral,
which is useful in feedback analysis, customer reviews, and social
media.
• Example Prompt Formula: “Classify the sentiment of this text:
[text].”
• Example: Classifying customer reviews for feedback.

Named Entity Recognition
Prompts
• Technique: Named Entity Recognition (NER) prompts extract
specific entities like names, locations, and dates from text,
enabling organized data extraction from large volumes of
information.
• Example Prompt Formula: “Identify and classify entities in:
[text].”
• Use Case: Entity extraction in legal documents or articles.

Text Classification Prompts
• Purpose: Text classification prompts categorize content into
predefined labels or themes, commonly used in organizing large
datasets, such as sorting emails or categorizing news articles.
• Example Prompt Formula: “Classify this text into the following
categories: [categories].”
• Example: Categorizing news articles by subject.

Text Generation Prompts
• Description: Text generation prompts provide AI with key details
to produce unique, tailored content, commonly applied in creative
writing, storytelling, and structured content generation.
• Example Prompt Formula: “Generate a story with these
elements: [elements].”
• Example: Writing creative text like stories or essays.

Conclusion
• Summary: This presentation has covered a wide range of prompt
engineering techniques, emphasizing how different methods can
direct AI responses to suit user needs effectively.
• Importance of Experimentation: Trying different prompt
combinations for optimal results.

AICT Lecture 03 -Command Prompts in AI.pptx

More Related Content

Similar to AICT Lecture 03 -Command Prompts in AI.pptx (20)

Recently uploaded (20)

AICT Lecture 03 -Command Prompts in AI.pptx