Generative AI For Dummies

Introduction

Welcome to Generative AI For Dummies, a groundbreaking book that’s one of the first, to my knowledge, to be professionally produced with the assistance of Generative AI and crafted by an award-winning, best-selling author in collaboration with a premier publishing house. To be absolutely clear, I wrote this book using GenAI models to help with research and drafting only. As impressive as they are, GenAI models and applications fail at writing books to professional standards. However, they can be useful when used properly in a junior-level assistant capacity.

Many people on the editorial, technical, and AI sides at Wiley collaborated to make this book possible and to make sure it met the highest standards. It’s a team of incredibly talented people who were fearless in pioneering book production with Generative AI (GenAI) to learn, advance, and help others find success in the GenAI space. Also know that I have researched, tested, and used many of the techniques and methods you find here to write this book. It’s been a hard and trying trek for all of us through the unknown — not to mention dealing with the stubbornness in GenAI behaviors. But it all proved worth it in the end. I hope you think so, too!

Rather than drone on about the technical aspects, my intention is to help you move past the hype and common frustrations in working with GenAI to actually produce usable works that meet your specifications and goals. I hope you find this book an inspiration, a real-world example of what can be done with GenAI, and an exceedingly helpful guide to help you level up your skills, too.

If you feel some unease about AI in general and GenAI in particular, know that your gut reaction is common and not entirely unwarranted. This technology will most certainly change the nature of work and how your job is done. But also know that GenAI is not going to take jobs away from most people. Someone good at using GenAI will. Be that someone!

You can learn this. It’s not as hard as you think!

About This Book

Although you can find lots of content on various GenAI models, applications, and the basics in prompting on YouTube and in blogs, articles, social media, and elsewhere, this book uniquely reveals and explains advanced prompting and other AI methods to generate professional-level outputs no matter which GenAI model you’re using or what type of content you are working to create.

It's a comprehensive text on the nitty-gritty details you need to push past the basics, the hype, and the common frustrations to make GenAI tools work like you need them to, yet these are things almost any user can do. Be kind to yourself as you experiment along your path to success.

If you’re already experimenting or working with GenAI, you’ll find several ways in this book to leverage what you already know and new tips and techniques to incorporate into your efforts to get even more from it.

As with any For Dummies book, the goal is for you to quickly access the information you need, no matter where you jump in. With that in mind, here are a few conventions I used when writing the book:

Generative AI and GenAI are used interchangeably, though I primarily use the shorter GenAI throughout the book.

References to GPT-like models or ChatGPT-like chatbots may or may not mean that they’re technically similar to these models and tools since this book covers a wide range of Generative AI options. For example, competing models may or may not have Large Language Models (LLMs) as their foundation, as ChatGPT does, but are still referred to as similar here because their user interface and function closely resemble those of ChatGPT. In this way, you as a user can more easily compare and understand the various GenAI tools on the market without getting dragged through the technical weeds.

Whenever I introduce a new GenAI-related term (and there are lots of them in this ever-changing space), I place it in italics upon its first mention in a chapter and follow up with a quick definition. Keywords and action steps appear in bold.

Some web addresses break across two lines of text. If you’re reading this book in print and want to visit one of these web pages, simply type the address exactly as it’s noted in the text, pretending the line break doesn’t exist. If you’re reading this as an ebook, you have it easy; just click the web address to be taken directly to the web page.

Foolish Assumptions

This book is for anyone seeking to understand, use, and improve their work with GenAI models and tools. Having said that, I did make a few assumptions about you, dear reader, as a practical matter:

You have at least a basic level of comfort and skill in working with computing devices, browsers, and web applications.

You possess a limited understanding of the full range of GenAI options and their capabilities.

You, like many users, are frustrated or stuck with GenAI outputs that fall far short of ideal in terms of meeting your goals and needs. Of the things that might make you seriously contemplate screaming out your window or would make a saint take to drinking, GenAI would be near the top of the list.

You are smart and pressed for time and, therefore, want all meat and no fluff in a fast and easy read. I hope I’ve hit that mark for you.

Icons Used in This Book

Occasionally you’ll come across some icons in the margins of this book that draw your attention to certain bits of the text. Here are what these symbols indicate:

Tip This icon points to tips and tricks you may want to use to make your work with ChatGPT easier, faster, more efficient, or simply more fun.

Remember This icon highlights information of particular importance in successfully understanding or using ChatGPT.

Warning This icon warns you of a stumbling block or danger that may not be obvious to you until it’s too late. Please make careful note of warnings.

Beyond the Book

In addition to the material in the print or ebook you’re reading right now, this product also comes with an access-anywhere Cheat Sheet. To access the Cheat Sheet, go to www.dummies.com and type Generative AI For Dummies Cheat Sheet in the search box. You’ll find helpful user tips and info on GenAI in its many forms, pointers on advanced prompt writing, and guidance on how to make it deliver the output you need precisely the way you need it, all within a convenient online article you can access on the go.

Where to Go from Here

This is a reference book, so you don’t have to read it cover to cover unless you want to. Feel free to read the chapters in any order and skip over any ones that don’t pertain to you or your needs. Each chapter is designed to stand alone, meaning you don’t have to know the material in previous chapters to understand the chapter you’re reading. Start anywhere and finish when you feel you have all the information you need for whatever task is at hand.

However, if your aim is to get better results from Generative AI immediately, you should read the chapters that specifically address prompting and the type of content you are trying to create, such as Chapter 2 for an introduction to the art of prompt engineering, Chapter 9 for many advanced prompting techniques and AI tactics, and chapters 7–10 for guidance on producing specific types of content. Or if you’re worried that AI will take your job or abuse your private information in some way, there’s good news for most of us in chapters 15 and 16.

Feel free to experiment with each new tool or tip you learn about in this book as you go. Many find it easy to follow along this way. But however you choose to learn and experiment with GenAI, you’ll likely find these tips and methods relatively easy to do. The hardest part is stretching your own imagination to allow yourself to reach further with each new project.

The key to success, however, lies in your own talent. GenAI tools are nothing without it.

Part 1

Diving into Generative AI Fundamentals

IN THIS PART …

Grasping the basics in using GenAI as a creative tool

Knowing the difference between generating and creating

Navigating through a variety of GenAI models

Layering GenAI to improve your work

Busting myths about GenAI

Practical considerations for working with GenAI

Chapter 1

Mapping the Lay of the Generative AI Land

IN THIS CHAPTER

Bullet Selecting the right GenAI model

Bullet Understanding GenAI capabilities

Bullet Creating with GenAI

Bullet Grasping its many uses

Bullet Conquering fears and piloting opportunities

Welcome to the exciting world of Generative AI (GenAI)! This chapter is your starting point in understanding the vast landscape of GenAI and its transformative capabilities. Whether you’re a curious beginner or a tech enthusiast, you’ll find the information here to be an accessible guide to the basics of GenAI. You can easily build on these skills through practice, regular use of an AI application, or by returning to this book from time to time to enhance your skills further.

So, What Exactly Is Generative AI?

You can think of AI (short for artificial intelligence) as incredibly sophisticated software. Although it doesn’t behave like any other software ever made, it is still software. Illustrations depicting AI as robots reflect the difficulty in drawing AI software in a way everyone will instantly recognize. But the robot is actually mindless hardware, and the AI is the smart brain-mimicking software installed to enable it to function in ways we consider to be intelligent in a nonorganic sense.

Technically speaking, GenAI refers to a subset of artificial intelligence technologies that use sophisticated natural language processing (NLP), neural networks, and machine learning (ML) models to generate unique and humanlike content. It belongs to a classification of AI called Large Language Models (LLMs), which analyze huge amounts of data in numerous languages including human languages, computer code, math equations, and images.

LLMs typically have a substantial number of parameters, which are numerical values used to assign weight and define connections between nodes and layers in the neural network architecture. Parameters can be adjusted to change the weights of various values, which in turn, changes what the model prioritizes in the prompt and data and how it interprets various data points, words, and connections.

Imagine you have a recipe for making a cake, and the recipe is your GenAI model. The ingredients — like flour, sugar, eggs, and butter — are like the data points, words, and connections in the model. Now, the amount of each ingredient you use (how many cups of flour, how much sugar, and so on) are like the weights of various values in the GenAI model or GenAI application.

Just as you might adjust the ingredients in your cake to make it sweeter or fluffier by adding more sugar or an extra egg, you can adjust the parameters in a GenAI model to change what it focuses on and how it interprets the information it’s given. If you want your GenAI to pay more attention to certain words or data points, you increase their weight just like adding more chocolate chips to your cake if you want it to be extra chocolatey. This way, the GenAI model, like your cake, turns out the way you want it to, based on what you prioritize in the recipe.

LLMs use parameters to predict the next word in a sequence — meaning they predict the word most likely to follow the words in your prompt, and then the word that most likely follows its first predicted word, and so on until the model believes it has finished the most probable pattern. It generates images in much the same way by predicting the image that follows your description in the prompt. The models can complete the process incredibly quickly. For example, LLMs like GPT-3 and GPT-4o developed by OpenAI are capable of processing billions of words per second. It is the speed of its response, the appearance of nuanced understanding, and its fluid use of natural language that gives GenAI interactions a humanlike feel.

Remember However, GenAI and LLMs are not human and do not think — again, they predict. It’s a very complicated prediction process, to be sure. Nonetheless, it is a prediction. And if anything happens to tilt its predictive capabilities, nonsense ensues. You can see one example of that in Figure 1-1, which is an OpenAI incident report about an adjustment they made to the model resulting in ChatGPT responding to users in incomprehensible gibberish.

Source: OpenAI incident report

FIGURE 1-1: A routine effort to optimize ChatGPT resulted in its producing gibberish in response to users’ prompts.

GenAI VERSUS VIRTUAL ASSISTANTS

AI models and applications are the software driving the robot or the autonomous car or whatever form it’s given in the corporeal world. But strictly speaking, AI has a digital form. Because of that, it can be squeezed into almost anything, and many a vendor does exactly that. You’ll find various types of AI are embedded or otherwise at use in all sorts of products and services. However, not all AI is the same.

Here are the main differences between GenAI apps like ChatGPT and virtual assistants like Siri, Alexa, and Google Assistant.

Virtual assistants:

This class of AI runs on a proprietary mix of technologies in a blend developed by their respective corporate owners. Certain components, such as machine learning, deep learning, natural language processing, smart search or search engines, and speech synthesis make the assistants appear and sound much like ChatGPT.

However, their responses are more limited than GenAI models. People typically use these to retrieve answers to common questions or perform uncomplicated tasks like where is the nearest pharmacy? or play a song by Taylor Swift rather than to generate original answers.

GenAI models (specifically ChatGPT in this comparison):

This class of AI runs on a single AI model, meaning on one version or another of Generative Pre-trained Transformers (GPT) AI models. GenAI is a broad category of AI that includes models capable of varying capabilities such as generating text, images, or computer code or some combination of these.

People typically use GenAI web apps, but some mobile apps and a few wearable devices are available as well. But in all cases, the apps run on a single GenAI model.

Unveiling the BIG Secret to Working Successfully with GenAI

If you remember nothing else I’ve written in this book, you must remember what I tell you in this section. For here is the big secret — the master key — that you need to make GenAI models work at the level you need them to perform. If you don’t grasp this, GenAI will likely appear to you to be nothing more than a fascinating toy or a tool that falls far too short of your expectations.

Tip In a nutshell, GenAI generates outputs that appear to be original thoughts or images from a computer, rather than results produced by very advanced, contextual predictive software. GenAI retrieves words or images pulled from a database and repurposes them into a new response. The big secret is that the humanlike feel in the conversation is an illusion. You are not having a conversation with a machine. It doesn’t understand a word you wrote in your prompt.

Remember Current GenAI models don’t think or create things per se, but instead generate new things from parts of old things found in its database. (The term things in this context being images, videos, numbers, or text, depending on the GenAI application you are using.) A GenAI output is the model’s best prediction of what you are seeking. In an oversimplified explanation of a complex technology, GenAI seeks to complete a pattern that you began with your prompt, which is your question or command as entered into the prompt bar on the GenAI’s user interface (UI). In other words, GenAI predicts what letters, words, or images are likely to follow those that are in your prompt. Its predictions are based upon comparison to patterns that exist within its training dataset and/or datasets to which it was subsequently given access.

Think of GenAI outputs as the result of repurposing or remixing information that the model has access to in datasets, including the following:

Data it is exposed to in its training database along with any additional data provided in subsequent fine-tuning.

Data added in system messages or prompts.

Data added via methods such as retrieval-augmented generation (RAG), which is a tactic to enhance accuracy, relevancy, and reliability by adding external sources to the GenAI’s database.

RAG combines the strengths of both information retrieval AI, which is a set of algorithms that retrieve contextually relevant information from huge datasets, and GenAI, which uses neural networks and machine learning models to generate new content. It might help to think of RAG as GenAI that is augmented by more traditional information retrieval AI, or retrieval AI for short.

Warning Since GenAI generates outputs that are the result of its remixing or repurposing of information, it has no concept of true or false, fact or fiction. GenAI can accurately define these terms, but it does not understand their meaning. It doesn’t understand anything you wrote in the prompt or that it wrote in its response. It only appears to understand terms and concepts. This is an illusion. This is why you must always factcheck its work.

Remember GenAI responses are limited to the confines of the data it has access to. Put another way, if its training data were a mound of Legos and there were no end caps in that mound, GenAI would build its outputs without end caps. It would not know that end caps exist at all. In the same way, it does not know fact from fiction unless those labels are applied to specific data points in its dataset. But, if a falsehood is labeled as fact, GenAI will unquestionably accept it as fact. It still doesn’t understand the difference.

To illustrate this analogy, I wrote a caption first and then used it as a prompt in Azure OpenAI Studio DALL-E playground (Preview). The result is the stunning concept illustration you see here in Figure 1-2.

Art generated by Azure AI Studio DALL-E playground.

FIGURE 1-2: If data were Legos, GenAI could only build things with the Lego pieces it has access to, and it is completely unaware that any other types of Lego pieces exist.

GenAI can repurpose and remix only the data it has access to, which is a major reason why GenAI outputs can be highly reliable or totally false or something in-between. The data itself can be insufficient — in one way or another — to provide the foundation or elements for the model to generate an accurate answer. Outdated data from an aging training dataset and data limited to too few perspectives or examples are common issues, but there are many others.

When — not if — outputs are wrong, people call them hallucinations. It’s unclear why no one calls them lies, falsehoods, or simply errors, but in any case, you cannot assume that GenAI outputs are solid enough to bet your life or business on without doing some serious factchecking first.

Remember While GenAI does consider context when it analyzes the words in your prompt, it does not understand you or what you said in the prompt. This is why you must not confuse GenAI with General AI, also known as Artificial General Intelligence (AGI). AGI does not yet exist outside of science fiction movies, books, and TV shows. Yet some people are so in awe of GenAI capabilities that they are sure this must be it — the thing from the movies that’s going to take over the world! This is not that.

Understanding the Infamous Finger Problem and Other GenAI Quirks

Perhaps the most wondrous thing about using GenAI is the delicate dance between human and machine that begets something neither would have made alone. But once you move past the first exhilarating moments of viewing GenAI marvels, you’ll begin to see a few cracks here and there.

For example, it is common for GenAI models to draw people with six or more fingers on one hand. This is typically because the patterns it sees in its data is of multiple fingers on one human hand. No clear pattern emerges of there being just five fingers on one hand, so GenAI can’t predict how many fingers it needs to generate.

Essentially GenAI is parroting the answer from its database. It doesn’t understand the question or the answer; therefore, it does not know to draw only five fingers. Instead, it looks for patterns in hands depicted by images or text in the datasets to which it has access. But the pattern of the total number of fingers is unclear. Images in most databases that GenAI models use typically show hands in different positions wherein only some fingers are visible or fingers from two hands or more are intertwined. GenAI cannot therefore see a consistent pattern of the total number of fingers per hand. However, if you were to ask the model how many fingers are on a hand, it will almost always tell you that there are five. Even though it gives you the right answer as to the number of fingers, it does not understand its own reply and, therefore, still doesn’t know the answer.

Data pattern inconsistency and the resulting probability prediction error is why you can end up with too many or too few fingers in any image GenAI generates. This is often the reason for other issues in images and videos that GenAI creates such as errors in shadowing or movement.

Remember Although GenAI is impressive, its reasoning is limited. In fact, it’s extremely difficult for GenAI to reason at all. To overcome this shortfall and make it more powerful, add one or more humans to the mix and you’ll soon see real magic in the result. It is the collaboration between you and this extremely sophisticated software that will take you to the goals you seek.

Figuring Out How to Work with GenAI — It’s All About Your Prompts

Here’s the thing: Natural human language is a computer language now. In the case of GenAI, this means that the machine still works like a machine and the human like a human, but they can now interact through a computer language that everyday, non-programmer types of people can understand and use.

However, you, the human, still must think like a machine to get the most out of GenAI. Ask any computer programmer how important it is to think like a machine while programming — and this is true regardless of their choice of programming language, be that JavaScript, Java, HTML/CSS, SQL, Python, English, or French.

And why is changing how you think important? Because you are not having a conversation with GenAI. You are giving instructions (and, yes, even when your prompt is a question, it is an instruction) on what you want the model to produce, much like any programmer does. You must think beyond the language to the depths of the result you seek. The value of a programmer is not their computer language knowledge, although that is important too, but the problem-solving ability that they can then convert into language that renders the precise solution the programmer wants to produce. This is how you need to think and work with GenAI models, too.

Your prompts need to be more concise and detailed than the typical conversations you have with another human. For one thing, you cannot make assumptions that a listener will automatically fill in common details because GenAI often doesn’t know those details. Despite appearances, GenAI does not think and doesn’t truly understand your prompt; many of the natural assumptions you make in speaking to another human will not work in the same way in interactions with these models.

Tip The GenAI winning formula: Machine speaks like a human. Human thinks like a machine. The better you get at telling GenAI what you want, the better it’ll get at giving you what you need. It’s all about practice.

Why GenAI appears so human

From crafting sentences to conjuring up images, composing music, or creating synthetic data, GenAI is a master in making something instantly that can often readily pass as human made.

The interesting thing is that its outputs are human made in some sense. GenAI can be thought of as a creative tool like an artist’s paints, crayons, and pencils. Those items produce images first imagined in the artist’s mind and executed by the artist’s hand and skill. Similarly, GenAI delivers outputs according to the user’s vision and skilled prompting.

Further, much of the data that GenAI models learn from is generated by humans. But it learns far more from this information than you might imagine. It also learns the habits, attitudes, biases, and other human attributes behind the text, audio, and image data that it consumes.

For example, GenAI models have been known to be lazy in the summertime around peak vacation periods — meaning GenAI models may produce less content in response to a prompt than usual. GenAI may even tell a user to get the information for themselves. Sometimes a GenAI model also responds slower than normal, announces a delay, or makes excuses.

Such actions aren’t due to a bug or a flaw in the system. The AI is merely mimicking human behavior. Models learn human behavioral patterns along with data patterns from their training dataset. They make no distinctions in the values of the information versus the behavior and so are likely to distribute both, or either, in their outputs.

GenAI can also deliberately lie and act angry or sad or cheerful for the same reason. It may even appear to ignore you from time to time. On the flipside, GenAI models tend to perform better when given a virtual reward or a compliment. Again, all of this is just mimicry of the human behaviors it has learned. It’s important to remain aware of such idiosyncrasies when using GenAI. Strategically playing into these GenAI quirks can level up the responses you pull from it.

Depending on the model you’re using, inputs and outputs can be in text, images, and/or audio forms. Unlike traditional AI, which analyzes, makes decisions, and delivers outputs drawn from data, Generative AI can repurpose information to create seemingly original outputs in a conversational or artistic manner. But it can also plagiarize and pillage the works of other humans. You must always check its outputs for grievous and potentially liable or dangerous behaviors.

Warning A key point to remember is that if you use GenAI you are legally liable for what it does. It is not a separate legal entity … responsible for its own actions as Air Canada once argued trying to defend itself in a court case after ChatGPT gave one of its customers incorrect information. You’ll find more information in the discussion on responsibilities in Chapter 3. (And if you’re curious about the Air Canada story, you can access it here: www.bbc.com/travel/article/20240222-air-canada-chatbot-misinformation-what-travellers-should-know)

Realizing the human influences behind Generative AI's abilities

It's important to distinguish between generating and creating. GenAI creates text or images by generating a response from repurposed information based on its prediction of the best match to your prompt. GenAI does not create in the truest sense, which dictionary.com defines as to cause to come into being, as something unique that would not naturally evolve or that is not made by ordinary processes.

GenAI works mostly by making predictions, which technically is an ordinary process in that predictions are a common thing that people and analytics do. However, GenAI’s prediction processes are quite extraordinary in that they exist at a level never before achieved. By machine standards this achievement is extraordinary because it generates a new response as opposed to a regurgitated response or picking one of a limited number of canned responses. By human standards, GenAI’s performance is extraordinary because it can analyze huge amounts of data and respond in a conversational manner or with a newly generated image in seconds or minutes.

Tip One way to remember the difference is to think To generate is AI, to create is human or human and AI.

But make no mistake, GenAI is not as humanlike as it appears. Another distinction is in motivation. Humans are motivated to create; some even feel driven by their passions. By contrast, GenAI is not motivated to generate anything. Ever. It doesn’t get hungry, thirsty, lonely, inspired, emotional, cold, hot, uncomfortable, dedicated to a cause, politically activated, or otherwise stimulated so there’s no reason for it to do anything at all.

You must provide the vision, the passion, and the impetus in a prompt. Then it will try to generate whatever that is for you. Otherwise, it will sit idle for centuries — or however long its supporting