Chapter 10

Flashcards

Natural Language Processing is the field that

deals with making computers perform tasks
that involve dealing with human language.

Some applications of Natural Language

Processing include spam filtering,
conversational agents, question answering,
information extraction, sentiment analysis, and
machine translation.

Conversational agents contain technologies

such as speech recognition, language analysis,
dialogue processing, information retrieval, and
text to speech.

Sentiment analysis involves determining the

sentiment or emotion expressed in a piece of text,
often used to analyze customer feedback or
social media posts.

Machine translation refers to the automatic

translation of text from one language to another
using artificial intelligence systems.

Large language models are sophisticated artificial

intelligence systems designed to understand and
produce natural texts in a way that is very similar
to the way humans communicate.

Large language models diDer from traditional

natural language models in terms of data size,
computing power, and applications and uses.
Generative Artificial Intelligence is a type of artificial
intelligence technology aimed at generating new
content, such as text, images, and video.

Some benefits of Generative Artificial Intelligence

include assisting in improving the quality of machine
translation, producing instant text translation, and
creating large language models.

Some challenges and risks of using Generative

Artificial Intelligence include high energy
consumption, potential biases in model outputs,
privacy violations, intellectual property rights issues,
and security vulnerabilities.

The risks of Generative Artificial Intelligence can be

mitigated by implementing best practices in designing and
engineering inputs, treating model outputs as preliminary
drafts needing review and auditing, using diverse and
comprehensive training datasets, and continuously
monitoring and evaluating model outputs.

Prompt Engineering involves providing specific

instructions or input to artificial intelligence
language models to ensure desirable outputs,
such as generating text, summaries, or responses.

Some practical examples of Prompt Engineering

include writing an email to a client, promoting a
product on social media, and summarizing texts
from files.

Generative AI Applications refer to the various

uses and implementations of Generative Artificial
Intelligence, including text generation,
summarization, translation, and content creation.
 Smart Notes
•Natural Language Processing (NLP)

•Natural languages are languages humans use to communicate, such as English and Arabic.

•NLP's goal is to make computers perform tasks related to human language, enable

communication between humans and machines, or improve human communication.

•Applications include spam filtering, conversational agents, question answering, information

extraction, sentiment analysis, and machine translation.

•Conversational agents contain technologies such as speech recognition, language analysis,

dialogue processing, information retrieval, and text to speech.

•Question answering systems respond to specific questions and are exemplified by IBM's

Watson winning the Jeopardy game show in 2011.

•Large Language Models (LLM)

•LLMs are models that process and generate large amounts of natural language data.

•They are different from traditional natural language models.

•Generative Artificial Intelligence (Generative AI)

•Generative AI is capable of creating new content rather than simply analyzing existing data.

•Benefits, challenges, and risks of using Generative AI are associated with prominent

companies developing it.

Bag of Words Model

•In language processing, vectors x are derived from textual data to reflect various linguistic

properties of the text.

•Feature extraction or feature encoding is a process of extracting features with text data

called the bag-of-words model.

•A language model is a probability distribution over entire sentences or texts.

•The bag-of-words (BOW) model turns arbitrary text into fixed-length vectors by counting

how many times each word appears.

•Vectorization is the process of turning arbitrary text into fixed-length vectors to make it

machine-readable.

•The first step of vectorizing documents is to determine the vocabulary, which is the set of

all words found in the document set.

•Word count involves counting how many times each word appears in the document set to

vectorize the documents.

•Length-6 vectors are created for each document using the bag-of-words model to

represent word counts in each text document.

•The bag-of-words model loses contextual information and only tells you what words occur

in a document, not where they occurred.

•Text classification is the task of assigning a set of predefined categories to free text.

•Text classifiers can be used to organize and categorize almost anything, such as news

articles by topics and user reviews by sentiments and opinions.

Large Language Model

•Large language models are sophisticated artificial intelligence systems designed to

understand and produce natural texts.

•These models are built on deep learning techniques and trained on huge datasets

containing texts from various sources.

•They can generate new texts, answer questions, translate between languages, and perform

other linguistic tasks with high efficiency.

•Large language models determine the probability of a certain string of words being present

in a sentence, and are considered large when they consist of a large number of parameters.

•They fall under the group of models known as foundation models, trained on different data

and adapted to perform a wide range of tasks.

•Large language models differ from traditional natural language models in data size and
computing power, as they are trained on huge datasets and require enormous

computational resources.

•They are used for machine translation, automatic summarization, question answering, and

creative content generation.

•Creating and developing large linguistic models involves three main steps: data collection
and processing, model training using deep learning techniques, and testing and evaluating

the model on unseen data.

•There are questions about whether the results generated using large language models can

be trusted.

•Generative Artificial Intelligence is defined as a type of artificial intelligence technology

aimed at generating new content.

Generative Artificial Intelligence

•Generative Artificial Intelligence has the ability to generate diverse and unlimited results
beyond the training data and relies on advanced technologies such as Deep Neural

Networks and Large Language Models.

•Relationship between the field of artificial intelligence and other types of machine learning:
Generative AI models have various capabilities such as generating new images, enhancing
and editing images and videos, composing musical pieces, summarizing, editing, and

critiquing texts, conducting conversations in natural language, translating texts and

speeches, and writing or generating programming code.

•Difference between Generative AI models and other AI models: Generative AI focuses on

creating new data similar to the data it was trained on, while other AI models focus on

understanding, classifying existing data, or making decisions based on the data. Generative
AI models have wide applications in art, design, content creation, entertainment, scientific
research, as well as benefits such as increased productivity, improved services, facilitating

research and analysis, reducing costs, and enhancing innovation.

•Benefits of using Generative Artificial Intelligence: Increased productivity, improving

services, facilitating research and analysis, reducing costs, and enhancing innovation.

•Challenges of creating generative artificial intelligence applications: Difficulty in collecting,

cleaning, and preparing large and diverse datasets, which can be a costly process.

•Proposed solutions: Utilizing synthetic data and activating partnerships for data exchange.
Risks and Solutions for Generative Artificial Intelligence

•The need for immense computing resources to train models can be expensive and

consume a lot of energy.

•Using data centers powered by renewable energy.

•Implementing interpretability and transparency techniques that allow for the explanation

and understanding of the model's decisions.

•The model's inability to explain how it produces any particular answer due to its reliance

on neural networks that contain billions of parameters.

•Need for specialized expertise in fields such as data science, programming, etc.

•The difficulty of attracting AI specialists due to their rarity and the high cost of hiring them.

•The absence of regulatory laws and policies related to privacy, intellectual property rights,

and the ethical use of technology that complicates effective governance of the field.

•Proposed solutions: learning best practices in designing and engineering inputs, treating

model outputs as preliminary drafts needing review and auditing, using retrieval-enhanced
methods, using a diverse and comprehensive training dataset, fine-tuning models, and

improving their outputs, continuously monitoring and evaluating model outputs.

•Educating users about protecting private or sensitive data.

•Implementing best security practices in development and operations.

•Conducting security tests and launching bounty programs to discover vulnerabilities and

address security loopholes.

•Risks of Generative Artificial Intelligence include hallucination, bias, privacy violation,

intellectual property rights, and security vulnerabilities.

•The most prominent applications of generative AI include speech.

Artificial Intelligence Models and Prompt Engineering

•Generative AI Models:

•Can be open source or accessed through API Key

•Applications include generating text, image, audio, and video content

•GAI Model is a pretrained model trained on diverse data

•Input and Output:

•Types of inputs include articles, brief descriptions, questions, sounds, music, speech,

images, drawings, video, scenarios

•Prompt Engineering:

•Defining clear goals for conversation

•Ensuring specificity and context

•Using clear and specific language

•Importance of Prompt Engineering:

•Clarifying context and providing relevant context is essential for best answers
Use of HTML Tags in Notes

•The given text demonstrates the use of HTML tags to distinguish the structure of the

document and organize the content.

•The text presents various prompts for guiding a generative AI to provide specific responses

or complete tasks.

•Prompts include role-play, using symbols, determining output format, providing references,

specifying steps, and providing examples.

•The examples provided in the text cover a range of practical tasks such as writing emails,

promoting products on social media, and using generative AI in the workplace.

•The use of tags such as [ ] indicates that specific text portions must be filled in and altered

by the user to customize the prompts.

Generative Artificial Intelligence in Education

•Course Title: Introduction to Artificial Intelligence

•Module 1: What is Artificial Intelligence?

•Definition of AI: Discuss the broad definition of AI as a branch of computer science that
aims to create systems capable of performing tasks that would typically require human

intelligence.

•History and Evolution of AI: Overview of the history from Alan Turing’s ideas to modern AI,

highlighting key milestones like the development of neural networks and the victory of

Deep Blue and AlphaGo.

•Types of AI: Explanation of artificial narrow intelligence (ANI), artificial general intelligence

(AGI), and artificial superintelligence (ASI).

•Module 2: Foundations of AI
•Machine Learning: Introduce the concept of machines learning from data, including

supervised, unsupervised, and reinforcement learning.

•Deep Learning: Dive into neural networks, emphasizing deep learning and its success in

tasks like image recognition and natural language processing.

•Algorithms and Techniques: Basic introduction to algorithms such as decision trees, support

vector machines, and clustering.

•Module 4: Hands-On AI Projects

•Tools and Libraries: Introduction to popular AI tools and libraries like TensorFlow, PyTorch,

and Scikit-learn.

•Simple AI Project: Guide students through a basic project, such as building a machine

learning model to predict house prices or classify images.

•Capstone Project: Students propose and develop a project that applies AI techniques to a

problem of interest in groups or individually.

•Exclusive mountain retreat located high above the clouds

•Description of the design and amenities of each cabin

•Emphasis on luxury and sustainability in the design

•Overall immersive experience with stunning panoramic views of the mountain range
Global Cloud Computing Overview

•Overview of the state of cloud computing globally

•Explores benefits, challenges, and experiences of various countries

•Defines cloud computing and highlights different types and services

•Discusses main benefits such as service elasticity, business continuity, cost reduction,

reliability, and security

•Addresses key challenges and potential solutions

•Includes global experiences from leading cloud computing countries

•Focuses on how these countries adopted cloud computing, challenges faced, and solutions

applied

•Emphasizes the importance of maintaining high standards of security and privacy

•Encourages the development of human resources and infrastructure to support this digital

transition
 Summary
"Advanced Applications of Machine Learning and AI in NLP and Generative AI"

This lecture covers Machine Learning Part(III), which includes Natural Language Processing

(NLP), Large Language Models (LLM), and Generative Artificial Intelligence (Generative AI).
NLP applications include spam filtering, conversational agents, question answering,
information extraction, sentiment analysis, and machine translation. LLMs are advanced AI
models that can understand and generate human language. Generative AI can create new

content and has applications in various industries, such as media and entertainment.

"Utilizing Bag of Words for Text Classification: A Simple Yet Effective Method"

The Bag of Words model is a feature extraction method that converts text into fixed-length
vectors by counting word occurrences. Despite losing contextual information, it's a simple
way to vectorize textual data. Text classification uses text classifiers to assign predefined

categories to free text, organizing and categorizing various types of content.

"Unleashing the Power of Large Language Models and Generative AI"

Large language models are AI systems trained on vast text datasets, using deep learning
techniques, to understand and produce human-like text. They differ from traditional natural

language models in terms of data size and computational power. Large language models
can be used for various tasks such as machine translation, automatic summarization,

question answering, and creative content generation. Generative Artificial Intelligence is a

type of AI technology focused on generating new content.

Unleashing the Potential of Generative Artificial Intelligence

Generative Artificial Intelligence (AI) is a technology that can generate diverse and unlimited
results beyond its training data, using advanced technologies like Deep Neural Networks

and Large Language Models. It has various applications, such as generating new images,
enhancing and editing media content, composing music, summarizing texts, and conducting

natural language conversations. Generative AI models differ from other AI models as they
focus on creating new data rather than understanding or classifying existing data. The
benefits of using Generative AI include increased productivity, improved services, facilitated
research, reduced costs, and enhanced innovation. However, challenges in creating

generative AI applications involve data cleaning and preparation, utilizing synthetic data,

and activating partnerships for data exchange.

The Future of AI: Advancements, Risks, and Proposed Solutions

AI in the future include speech synthesis and text-to-image generation. However, these
advancements also pose risks such as hallucination bias, privacy violations, intellectual

property rights issues, and security vulnerabilities. To address these challenges, various
solutions are proposed such as improving data quality, developing interpretability and
transparency techniques, fine-tuning models, and establishing ethical guidelines and

regulatory policies.

"Prompt Engineering for Effective Generative AI Interaction"

Using Generative AI, a pretrained model processes diverse data to generate new content
based on user inputs. Inputs can include articles, descriptions, questions, sounds, speech,
images, drawings, videos, or scenarios. Prompt engineering is crucial for effective interaction

with generative AI, focusing on clarity, context, and creativity in commands.

"Exploring Diverse Topics through Writing"

1) Compare mobile phone brands, 2) Role-play as a historian discussing WWII, 3) Summarize

text using symbols, 4) Write product comparison table, 5) Use provided references on smart

cities, 6) Analyze and classify tweets, 7) Write expression of hope.

"Exploring the Impact of AI in Education and Design"

Chocolate Chip Cookie Ice Cream Sandwich now available at the bakery. Limited time offer.
AI uses in various fields, like education, preparing content for AI course. Course includes
introduction to AI, machine learning, deep learning, and AI projects. AI also applied in

design, creating images from text for exclusive mountain retreat concept.

"Global Perspectives on Cloud Computing: Benefits, Challenges, and Experiences"

This document discusses the state of cloud computing globally, covering benefits,
challenges, and experiences of various countries. It defines cloud computing, its types, and
services, then highlights main benefits and challenges. The document also shares global

experiences from Singapore, Finland, and Sweden, describing how they've used cloud
computing to enhance government and business services. It concludes by encouraging
other countries to adopt similar strategies while emphasizing security, privacy, and human

resources development.
Quiz
‫‪Good luck on your finals J‬‬

‫‪Made by: Rayan MohammadNaeem Tashkandy‬‬

‫ﻻ ﺗ ﻧ ﺳ و ﻧ ﺎ ﻣ ن ﺻ ﺎ ﻟﺢ د ﻋ ﺎ ﺋ ﻛ م ﷲ ﯾ و ﻓ ﻘ ﻛ م ﯾ ﺎ ر ب ا ﻣ ﯾ ن‬
‫ر اﺟ ﻌ و ا اﻟﺳ ﻼ ﯾد ا ت ﻛ ﻣ ﺎ ن ا ﺣ ﺗ ﯾ ﺎ ط‬