Unlocking the Secrets of Prompt Engineering: Master the art of creative language generation to accelerate your journey from novice to pro

Unlocking the Secrets of Prompt Engineering

Copyright © 2023 Packt Publishing

All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, without the prior written permission of the publisher, except in the case of brief quotations embedded in critical articles or reviews.

Every effort has been made in the preparation of this book to ensure the accuracy of the information presented. However, the information contained in this book is sold without warranty, either express or implied. Neither the author, nor Packt Publishing or its dealers and distributors, will be held liable for any damages caused or alleged to have been caused directly or indirectly by this book.

Packt Publishing has endeavored to provide trademark information about all of the companies and products mentioned in this book by the appropriate use of capitals. However, Packt Publishing cannot guarantee the accuracy of this information.

Group Product Manager: Niranjan Naikwadi

Publishing Product Manager: Nitin Nainani

Book Project Manager: Aishwarya Mohan

Senior Content Development Editor: Debolina Acharyya

Technical Editor: Reenish Kulshrestha

Copy Editor: Safis Editing

Proofreader: Safis Editing

Indexer: Manju Arasan

Production Designer: Jyoti Kadam

DevRel Marketing Coordinators: Vinishka Kalra

First published: January 2024

Production reference: 1111223

Published by Packt Publishing Ltd.

Grosvenor House

11 St Paul’s Square

Birmingham

B3 1RB, UK

ISBN 978-1-83508-383-3

www.packtpub.com

To my incredible wife, Maria Olga – you are my rock and my best friend. Thank you for always believing in me. To my son, Daniel, and daughter, Andrea – you both make me strive to be a better person every day. Your unwavering belief in me gave wings to this dream. This book is for you.

– Gilbert Mizrahi

Foreword

In the dynamic and fast-evolving landscape of AI (artificial intelligence), Gilbert Mizrahi’s latest book on the new discipline of Prompt Engineering stands as a beacon of practical knowledge and insight. As Founder and CEO of Aptima, Inc., I have known the author for more than 20 years since we were colleagues in the Company, inventing novel ways to engineer productive collaborations between humans and machines. It is my distinct honor and privilege to introduce this comprehensive and pioneering work.

Gilbert has always been at the forefront of innovation and technology since his days at Stanford, and this book is a testament to his deep understanding and insightful approach to large language models (LLMs) and generative AI. This work is not just a technical guide; it is a journey through the complex world of prompt engineering, underscored by a commitment to sound practices and a deep understanding of the transformative power of AI.

The book begins with a solid foundation in the basics of LLMs and moves through the intricacies of prompt engineering with the precision and clarity rarely found in the computer science and engineering field. His emphasis on ethics throughout the book is particularly commendable, ensuring that as we advance technologically, we do so with a conscientious mindset. As we experience throughout the book, prompt engineering is akin to crafting a meticulous recipe, where each ingredient – or, in this case, each element of the prompt – must be carefully honed to guide LLMs towards desired outcomes.

As we delve into the subsequent chapters of this incredible journey, the book masterfully decomposes and then synthesizes the core concepts. It also highlights innovative applications of prompt engineering across various industries. We are given a vivid glimpse into the future where prompt engineering could revolutionize fields like healthcare, offering compelling applications in clinical decision support, patient education, and drug discovery.

Gilbert’s foresight in anticipating future trends and breakthroughs in LLMs underscores the need for collaboration and continual learning in this field. His exploration of the diverse applications of LLMs illustrates their potential to fundamentally transform nearly every industry and domain. The book not only serves as a technical guide but also as a behavioral compass in the digital age, emphasizing the need for wisdom, care, and transparency as these capabilities advance.

In conclusion, Gilbert Mizrahi’s book is an indispensable resource for both newcomers and seasoned professionals in AI. It is a guide to understanding and mastering the art and science of prompt engineering to establish a fruitful dialog between the human user and the artificial intelligence, and a reminder of the collaborative effort required to harness its full potential responsibly. This book is a beacon for those navigating the evolving landscape of AI, reflecting a profound understanding of the technological, ethical, and practical aspects of LLMs.

I recommend this book to all who seek to explore the extraordinary possibilities of AI and prompt engineering. It is with great pride that I introduce this significant contribution to the field of AI, celebrating the achievements of a long-time colleague and friend.

Daniel Serfaty

Founder and CEO

Aptima, Inc.

Contributors

About the author

Gilbert Mizrahi is a product strategist, educator, and seasoned entrepreneur with a proven track record across a variety of industries. He brings a wealth of knowledge and experience in interactive data visualization, product strategy innovation, generative AI, and Software as a Service (SaaS). As a co-founder of Twnel, Gilbert steers product R&D, leveraging his expertise in artificial intelligence to craft cutting-edge solutions that enhance communication and productivity for businesses. His passion for data science and product growth is mirrored in his ventures.

He holds a master’s degree in operations research from Stanford University and a bachelor’s degree in industrial engineering. Gilbert's extensive background includes serving as a mentor at MassChallenge and Newchip Accelerator, business strategy and product development consultant at Looi Consulting, and holding senior research and technology positions at Aptima Inc.

Gilbert's forward-thinking approach, entrepreneurial spirit, and dedication to fostering innovation make him a vanguard in the application of generative AI and strategy in product development.

About the reviewers

Daniel Mizrahi, a software engineer at Google, specializes in automating third-party service deployments to Google Cloud. Over the past two years, he has focused on creating systems that not only enhance operational efficiency but also pave the way for more advanced cloud-based applications. His experience has given him a profound understanding of the nuances of modern cloud-based systems.

Before his current role at Google, Daniel was on the Amazon Prime Video team, working specifically on a portal for content providers. This experience honed his skills in creating user-focused software solutions and deepened his appreciation of the intricate relationship between software engineering and digital content distribution.

Daniel holds a master’s degree in computer science from the University of Southern California with a focus on machine learning and artificial intelligence. Daniel’s academic background forms the bedrock of his technical expertise.

Divit Gupta, a seasoned IT professional with 20 years of industry expertise, excels in driving strategic architecture initiatives and providing leadership in multi-pillar sales cycles. With a global impact, he spearheads technical partnerships, defines team vision, and champions new strategic endeavors.

As the host of popular podcasts such as Tech Talk with Divit, Live Labs with Divit, and Cloud Bites with Divit, he showcases Oracle’s technological initiatives and leadership. In 2022–23, he served as Oracle TV’s correspondent for Cloud World. His passion for knowledge sharing extends to international conference talks, technical blogs, and multiple books on emerging technologies.

A recognized expert, Divit presented on Oracle Database technology at Oracle CloudWorld FY 2023. Holding over 40 certifications from Microsoft, Oracle, AWS, and Databricks, he remains at the forefront of technology.

David Santiago Castillo, with an extensive tenure of over a decade at Twnel, is a seasoned software developer who has played a pivotal role in the evolution of the company’s communication platform. Twnel, initially founded as a messaging platform, has since transformed into a cutting-edge solution that automates business processes through conversational user interfaces. In the rapidly evolving landscape of AI and natural language processing, David has been at the forefront of harnessing the power of large language models to enhance and expand Twnel’s automation capabilities.

Table of Contents

Preface

Part 1: Introduction to Prompt Engineering

1

Understanding Prompting and Prompt Techniques

Technical requirements

Introducing LLM prompts

How LLM prompts work

Architecture

LLM training

A journey from prompt to reply – how inference helps LLMs fill in the blanks

Types of LLM prompts

Components of an LLM prompt

Adopt any persona – role prompting for tailored interactions

Few-shot learning – training models with example prompts

Finding your voice – defining personality in prompts

Using patterns to enhance prompt effectiveness

Mix and match – strategic combinations for enhanced prompts

Exploring LLM parameters

How to approach prompt engineering (experimentation)

The challenges and limitations of using LLM prompts

Summary

2

Generating Text with AI for Content Creation

Using AI for copywriting

Creating social media posts

Writing a Twitter thread

Writing an Instagram post

Producing high-converting sales copy

Writing video scripts

Generating blog posts, articles and news

Creating engaging content with AI

How to use AI for personalized messaging

Creating tailored content with AI

Summary

Part 2: Basic Prompt Engineering Techniques

3

Creating and Promoting a Podcast Using ChatGPT and Other Practical Examples

Crafting podcast questions for celebrity guests

Preparing podcast questions with everyday guests

Identify topics, ideas, and potential guest speakers for your podcast

Using AI to promote a podcast

Writing a summary of the podcast episode

Crafting engaging quotes for social media promotion

Conceptualizing podcast highlight reels

Repurposing podcasts into shareable blog content

Identifying insightful interview questions

Sharpening interview skills with AI-generated responses

Generating strategic questions for client engagements with AI

Summary

4

LLMs for Creative Writing

Using AI for creative writing

Using AI to generate fiction

Using AI to write poetry

Summary

5

Unlocking Insights from Unstructured Text – AI Techniques for Text Analysis

Sentiment analysis – AI techniques for emotion detection in text

Organizing unstructured data – using AI for automated text categorization and data classification

Cleaning up dirty data – how AI identifies and resolves issues in datasets

Making sense of unstructured data – pattern matching for information extraction

Summary

Part 3: Advanced Use Cases for Different Industries

6

Applications of LLMs in Education and Law

Creating course materials with ChatGPT

Creating handouts and other materials

Creating handouts for the unit

Creating solved examples

Word problems

Creating quizzes

Creating rubrics

Creating cloze comprehension tests

AI for legal research

Reviewing legal documents using an LLM

Drafting legal documents with an LLM

AI for legal education and training

LLMs for eDiscovery and litigation support

AI for intellectual property (IP) management

Other applications of LLMs for lawyers

Summary

7

The Rise of AI Pair Programmers – Teaming Up with Intelligent Assistants for Better Code

Code generation with coding assistants

From confusion to clarity – AI explains what code does in plain English

Commenting, formatting, and optimizing code

Fixing faulty code – how AI transforms the debugging process

Translating code from one language to another

Case study 1 – developing a website code using AI

Case study 2 – creating a Chrome extension using AI

Summary

8

AI for Chatbots

Technical requirements

How to use GPT-4 APIs and other LLM APIs to create chatbots

Building conversational interfaces with LLM APIs

How to use AI for customer support

Case study – a chatbot using AI to assist users in ordering products

Case study – creating interactive quizzes/assessments and deploying them as chatbot flows

Summary

9

Building Smarter Systems – Advanced LLM Integrations

Automating bulk prompting with spreadsheets

Integrating LLMs into your tech stack using Zapier and Make

Creating and translating product descriptions

Moving beyond APIs – building custom LLM pipelines with LangChain

LangChain’s building blocks

LangChain's no-code tools – Langflow and Flowise

LangSmith – debug, test, and monitor your LLM workflows

The future of LLM integration – plugins, agents, assistants, GPTs, and multimodal models

Summary

Part 4: Ethics, Limitations, and Future Developments

10

Generative AI – Emerging Issues at the Intersection of Ethics and Innovation

Exploring the ethical challenges of generative AI

Trust and accountability challenges of generative AI

Economic impact considerations

Environmental sustainability issues

Societal risks and reflections

Broader societal impacts

What machine creativity reveals about cognition

Concerns in defense and healthcare

The path forward – solutions and safeguards

Summary

11

Conclusion

Recap of the book’s content

Expanding possibilities – innovative prompt engineering applications

Achieving intended outcomes – prompt engineering goals

Understanding limitations and maintaining oversight

Summary

Index

Other Books You May Enjoy

Preface

The advent of large language models (LLMs), such as GPT-4, Bard, and Claude, represents a seismic shift in artificial intelligence (AI) capabilities. Fueled by vast datasets and compute power, these models can generate astonishingly human-like text and engage in complex dialogue. However, their potential is only fully realized through the art of prompt engineering – the process of carefully crafting the prompts that activate the models. Prompts encode instructions, context, examples, and guardrails to channel the models’ capabilities for specific tasks. Mastering prompt engineering unlocks the immense power of LLMs for a myriad of applications, from content creation to data analysis and beyond.

This book serves as a practical guide to prompt engineering across different domains. Through concrete examples and real-world case studies, readers will learn effective techniques for decomposing problems into discrete prompts, providing relevant background knowledge, iteratively refining prompts, and shaping model outputs. The journey equips readers with strategies to tap into the versatility of LLMs with properly engineered prompts. By following the techniques in this book, anyone can harness the problem-solving skills of AI systems such as GPT-4 and Claude to build solutions that create real value.

Who this book is for

Unlocking the Secrets of Prompt Engineering is written for anyone who wants to become an expert at crafting prompts for AI systems such as ChatGPT. Whether you’re a total beginner or have some experience in prompt engineering, this book will help you master the art and science of creating effective prompts. I’ve designed it for a broad audience, including students, researchers, entrepreneurs, marketers, customer service agents, and other professionals who want to utilize the power of prompts to get the most out of AI. My goal is to provide actionable strategies and techniques you can apply right away to improve your prompting skills. By the end of the book, you’ll know how to structure prompts that clearly communicate your intent, include the right amount of context and examples, and elicit the desired response from an AI system. The practical knowledge in this book will make you a prompt engineering pro!

What this book covers

Chapter 1, Understanding Prompting and Prompt Techniques

This introductory chapter provides a comprehensive overview of LLM prompts and the foundations of prompt engineering. It explores the components of prompts, different prompting techniques, LLM parameters, and a systematic framework for experimentation to craft effective prompts. The chapter also discusses challenges such as verbosity and inconsistency that need to be addressed. By equipping you with core knowledge about prompt engineering and how to guide LLM behavior, this chapter lays the groundwork for harnessing the power of AI for diverse applications in the following chapters.

Chapter 2, Craft Compelling Content Faster with AI Assistance

This chapter explores leveraging AI tools such as ChatGPT to generate, outline, and draft initial versions of content including social media posts, sales copy, video scripts, and articles. It covers providing context and examples to guide the AI, personalizing messaging, customizing the tone and voice, and refining the raw AI output. While AI shows promise to enhance human creativity and productivity in content creation through these techniques, human oversight remains critical. The key lessons focus on thoughtfully combining AI assistance with human creativity and intent to develop engaging and high-quality content.

Chapter 3, Creating and Promoting a Podcast Using ChatGPT and Other Practical Examples

This chapter provides practical examples of leveraging AI tools, for example, ChatGPT, for tasks such as crafting an engaging podcast and job interview questions. It explores prompts and techniques to identify podcast topics, potential guests, and promotional content. For job interviews, it covers how both interviewers and candidates can use AI to strategize relevant questions and thoughtful answers. The key lessons focus on using AI to accelerate preparation, idea generation, and content creation for podcasts and interviews while enhancing human creativity.

Chapter 4, LLMs for Creative Writing

This chapter explores how writers can leverage AI tools such as ChatGPT to enhance different aspects of the creative writing process. It provides examples of crafting prompts to generate ideas, characters, and plots for fiction as well as techniques for writing original poetry. The key lessons focus on using AI to spark imagination while retaining authorial vision and voice. With the right balance of human creativity and AI assistance, these models can accelerate idea generation, improve drafts through editing, and open new creative frontiers.

Chapter 5, Unlocking Insights from Unstructured Text - AI Techniques for Text Analysis

This chapter explores key applications of AI techniques such as sentiment analysis, data classification, data cleaning, and pattern matching to extract insights from unstructured text. It provides examples of using these techniques to perform tasks such as gauging emotion in content, categorizing data, resolving inconsistencies, and extracting structured information. The key lessons focus on leveraging AI to automate the analysis of qualitative data, saving time and effort while improving accuracy. With the right techniques, AI enables anyone to unlock value from the proliferation of unstructured text data.

Chapter 6, Applications of LLMs in Education and Law

This chapter demonstrates applications of AI systems such as ChatGPT in education and legal domains. It provides examples of using these tools to generate personalized course materials, practice questions, and rubrics tailored to learning objectives. For legal professionals, the chapter explores leveraging LLMs for research, drafting documents, intellectual property management, training law students, and other emerging use cases. However, human validation of AI responses remains critical. When thoughtfully implemented, tools such as ChatGPT show immense potential to assist professionals in education, law, and other fields by automating repetitive tasks and enhancing productivity.

Chapter 7, The Rise of AI Pair Programmers - Teaming Up with Intelligent Assistants for Better Code

LLMs such as GPT-4 are transforming coding by generating functional code blocks, explaining code, debugging, optimizing performance, and translating between programming languages. This chapter provides case studies demonstrating the use of AI to rapidly develop website code and Chrome extensions, allowing developers to focus on design rather than rote coding tasks. AI coding assistants such as GitHub Copilot leverage GPT-3 and GPT-4 to provide autonomous code generation tailored to developers’ needs. While AI can accelerate development, human oversight is still needed to review and refine the generated code before deployment. AI is unlikely to wholly replace developers soon, but it can augment human creativity and problem-solving abilities in coding. The future will involve fluent human-AI collaboration, with coders and assistants working together symbiotically.

Chapter 8, Conversational AI – Crafting Intelligent Chatbot LLMs

Chatbots powered by LLMs such as GPT-3/4 and Claude are transforming conversational AI and enabling more natural, human-like digital experiences. As demonstrated through the detailed examples in this chapter, these powerful generative models allow bots to truly understand natural language, hold free-flowing conversations with users, and complete sophisticated workflows from commerce transactions to personalized assessments.

The key to unlocking their capabilities is thoughtful prompt engineering. Developers can inject critical context, domain knowledge, business logic, data sources, and more into the prompts to shape the bot’s behavior. While interacting in the playground provides a glimpse of the potential, custom solutions built on LLM APIs open up many more possibilities.

Chapter 9, Building Smarter Systems – Advanced LLM Integrations

This chapter explored various techniques for integrating LLMs into practical workflows to unlock new possibilities. Easy-to-use templates such as SheetSmart simplify setting up formulas in spreadsheets to prompt LLMs such as GPT-3.5 in bulk. More powerful automation platforms such as Zapier and Make enable connecting web applications into pipelines with LLM APIs. This allows automating processes such as generating competitive intelligence briefings by ingesting data sources into an LLM.

For full customization, developer tools such as LangChain, Flowise, and Langflow provide frameworks for building sophisticated LLM applications involving reasoning, conversation, and contextual recommendations. The walk-throughs in this chapter demonstrate sample integrations for extracting insights from customer data to enrich CRM systems and conversing with PDF documents using LLMs.

Chapter 10, Generative AI – Emerging Issues at the Intersection of Ethics and Innovation

Generative AI introduces profound challenges around trust, accountability, bias risks, economic impacts ranging from productivity gains to job displacement, massive computational needs threatening sustainability, subtle societal risks, and philosophical questions around machine creativity. Solutions require collaboration on ethics by design, algorithmic assessments, thoughtful regulations, inclusive governance, and upholding human rights. The choices made today on AI ethics and governance will have profound implications. With humanism guiding development, these technologies can be steered toward uplifting and enriching society.

Chapter 11, Conclusion

Prompt engineering represents a breakthrough in guiding generative AI systems such as LLMs to automate tasks and enhance human capabilities across industries. Meticulously refining prompts based on outputs is key to steering these models, much like adjusting ingredients when cooking. For now, human expertise remains essential to oversee AI’s nascent abilities. Looking ahead, techniques such as conditional and causality prompting could enable more reliable, personalized applications. Healthcare and other fields, such as engineering and finance, exhibit immense potential for prompt engineering to assist professionals. However, we must acknowledge the current limitations and implement thoughtful governance to manage risks responsibly. Platforms integrating models with services and tools customizing outputs by training on unique data will expand the possibilities further. While focused on text generation, prompt engineering will grow even more versatile as multimodal LLMs advance.

Overall, this book provides an introductory survey of techniques to ignite exciting possibilities for transforming nearly every facet of life and work through thoughtful prompt engineering.

To get the most out of this book

To get the most out of this book on prompt engineering, you need access to OpenAI’s ChatGPT, Anthropic’s Claude 2, and Google’s Bard. We start from fundamentals, so no deep technical expertise is required. You should be keen to engage creatively and iteratively in crafting effective language model prompts. An inquiring mindset will be rewarded as prompt engineering develops skills in articulating tasks and unlocking capabilities. Our goal is to set you on a self-directed journey.

We primarily focus on crafting prompts for OpenAI’s ChatGPT, Claude from Anthropic, and Google’s new Bard language model, covering the available techniques across these industry-leading systems. The code and examples are compatible with macOS, Windows 10+, or common Linux distributions. Hardware requirements are modest – an internet connection and computer from the past 5-7 years will suffice.

We’ve structured the book to accelerate applying prompt engineering to real problems. If grasping concepts rapidly and jumping quickly into hands-on activities suits your learning style, you have found the right guide.

Download the example code files

There is a code file in ch 08 whose github link has been pasted below https://github.com/PacktPublishing/Unlocking-the-Secrets-of-Prompt-Engineering/tree/main

We also have other code bundles from our rich catalog of books and videos available at https://github.com/PacktPublishing/. Check them out!

Conventions used

There are a number of text conventions used throughout this book.

Code in text: Indicates code words in text, database table names, folder names, filenames, file extensions, pathnames, dummy URLs, user input, and Twitter handles. Here is an example: The only thing that changed was that browser-action changed to action in the manifest.

Prompts are written as follows:

For the first set of questions, provide for each three potential good answers

Responses to prompts are written as follows:

1. As a successful solopreneur, how important do you believe personal branding is in the landscape of solopreneurship, and how has your own personal brand contributed to your success?

2.. Do you believe there's a potential risk for solopreneurs in blurring the lines between personal and professional branding? How do you maintain a balance?

3. Personal branding seems to be a powerful tool for standing out from the competition. How do you ensure your personal brand reflects your unique value proposition to your clients?

4. For solopreneurs who are just starting, what would be your advice on establishing a strong and credible personal brand?

When we wish to draw your attention to a particular part of a code block (in this case, a prompt), the relevant lines or items are set in bold:

Use the [text] extracted from an invoice using OCR. Structure it as a JSON object with the structure shown in the [JSON model].

[text]=

TEXT GENERATED BY OCR

[JSON model] = (the model from above)

keep the same structure of the JSON as in the model, including the same keys. Output:

Get in touch

Feedback from our readers is always welcome.

General feedback: If you have questions about any aspect of this book, email us at [email protected] and mention the book title in the subject of your message.

Errata: Although we have taken every care to ensure the accuracy of our content, mistakes do happen. If you have found a mistake in this book, we would be grateful if you would report this to us. Please visit www.packtpub.com/support/errata and fill in the form.

Piracy: If you come across any illegal copies of our works in any form on the internet, we would be grateful if you would provide us with the location address or website name. Please contact us at [email protected] with a link to the material.

If you are interested in becoming an author: If there is a topic that you have expertise in and you are interested in either writing or contributing to a book, please visit authors.packtpub.com.

Share Your Thoughts

Once you’ve read Unlocking the Secrets of Prompt Engineering, we’d love to hear your thoughts! Please click here to go straight to the Amazon review pagefor this book and share your feedback.

Your review is important to us and the tech community and will help us make sure we’re delivering excellent quality content.

Download a free PDF copy of this book

Thanks for purchasing this book!

Do you like to read on the go but are unable to carry your print books everywhere?

Is your eBook purchase not compatible with the device of your choice?

Don’t worry, now with every Packt book you get a DRM-free PDF version of that book at no cost.

Read anywhere, any place, on any device. Search, copy, and paste code from