Unlock exclusive free benefits that come with your purchase, thoughtfully crafted to supercharge your learning journey and help you learn without limits.

Here’s a quick overview of what you get with this book:

Next-gen reader

Figure 1.1: Illustration of the next-gen Packt Reader’s features

Our web-based reader, designed to help you learn effectively, comes with the following features: Multi-device progress sync: Learn from any device with seamless progress sync. Highlighting and notetaking: Turn your reading into lasting knowledge. Bookmarking: Revisit your most important learnings anytime. Dark mode: Focus with minimal eye strain by switching to dark or sepia mode.

Interactive AI assistant (beta)

Figure 1.2: Illustration of Packt’s AI assistant

Our interactive AI assistant has been trained on the content of this book, to maximize your learning experience. It comes with the following features: Summarize it: Summarize key sections or an entire chapter. AI code explainers: In the next-gen Packt Reader, click the Explain button above each code block for AI-powered code explanations. Note: The AI assistant is part of next-gen Packt Reader and is still in beta.

DRM-free PDF or ePub version

Figure 1.3: Free PDF and ePub

Learn without limits with the following perks included with your purchase: Learn from anywhere with a DRM-free PDF copy of this book. Use your favorite e-reader to learn using a DRM-free ePub version of this book.

Unlock this book’s exclusive benefits now Scan this QR code or go to packtpub.com/unlock, then search for this book by name. Ensure it’s the correct edition.
Note: Keep your purchase invoice ready before you start.

AI is the grand umbrella term that encompasses all forms of computational systems that can perform tasks that normally require human intelligence. AI encompasses a wide range of subfields, including machine learning, deep learning, neural networks, Natural Language Processing (NLP), and robotics. Its applications are diverse, ranging from medical diagnosis and financial analysis to self-driving cars and virtual personal assistants.

The term artificial intelligence was first introduced by John McCarthy in 1956 during the Dartmouth Conference, marking the birth of AI as a field of study. AI gained momentum with the rise of machine learning, which focused on developing algorithms that allow computers to learn from data and make predictions or decisions without being explicitly programmed for specific tasks. The availability of large datasets and advances in computing power facilitated the development of more complex machine learning models.

The mid-2000s marked a significant breakthrough in AI with the advent of deep learning. Deep learning, a branch of machine learning, is characterized by neural networks with multiple layers. It began to gain prominence around 2006, largely driven by Geoffrey Hinton’s groundbreaking work in developing techniques that enabled AI systems to learn in a human-like manner. Deep learning models achieved remarkable success in tasks such as image and speech recognition, NLP, and gaming. This era’s progress was propelled by the availability of large datasets, powerful GPUs, and improved algorithms, all of which facilitated the training of increasingly complex models.

Generative AI, which can generate content that closely resembles human creation, saw significant advancements in 2014. It began with the capability to create realistic images from noise maps. Over time, it has evolved to craft an extensive variety of content, spanning from textual compositions and imagery to video clips, musical pieces, and synthesized speech.

The early 2020s were marked by an AI boom, particularly with the advancements in deep learning and the development of LLMs. These models are capable of summarizing, reading, or generating text in a manner similar to human communication, which has led to a substantial expansion of generative AI systems. Advanced chatbots such as ChatGPT, Copilot, and LLaMA have contributed greatly to the AI landscape, transforming our interaction with technology and unlocking unprecedented levels of efficiency and creative potential.

AI continues to advance at an unprecedented pace. However, its core components are deeply interconnected, starting with the broad foundation of AI and progressing to more specialized areas such as machine learning, deep learning, and, ultimately, specialized models such as generative AI and LLMs. The core components of AI and their relationships are outlined next, illustrating how each one is interconnected within the broader AI ecosystem:

• AI is the broad field – the “umbrella”
• Machine learning is a core component of AI – it’s a method within AI
• Deep learning is a specialized subcomponent of machine learning
• Generative AI is an application area (or functional branch) of deep learning
• LLMs are a specific type of generative AI – very specialized components built on top of deep learning architectures

The following diagram offers a visual guide to these core AI components:

Figure 1.4 – Visual guide illustrating the layers within AI systems

Note that this visual guide depicts the core components of AI in layers, with each component in an inner layer being a subset of the component in the outer layer. Each layer also builds upon the capabilities of the outer layer, illustrating how each foundational technology, such as machine learning, paved the way for more advanced developments, such as deep learning.

As AI continues to advance, its transformative impact is being felt across a wide range of industries. In healthcare, AI is revolutionizing the sector by helping doctors with diagnoses, creating personalized treatment plans, and accelerating the pace of drug discovery. Banks and financial institutions are leveraging AI’s power to detect fraudulent activities, execute algorithmic trades, and manage risks. In the automotive industry, AI is behind the wheel of self-driving cars and boosting safety with advanced driver-assistance systems. Retailers are tapping into AI to tailor customer recommendations, streamline inventory management, and automate client services. In the manufacturing sphere, AI is used to optimize supply chain logistics, enable predictive maintenance, and ensure quality control through cutting-edge analytics and robotic automation.

The applications of AI across industries highlight its transformative potential and the ongoing technological evolution. While AI-powered tools and systems have already made a significant impact, the ongoing research and development continue to overcome present limitations and unlock new possibilities, paving the way for innovations that could revolutionize industries and everyday life.

With AI continuing to evolve, understanding the technologies driving this transformation becomes crucial. In the next few sections, we will delve into the core AI components that have shaped modern AI. By gaining a clear understanding of how each of these technologies works, you’ll be better prepared to use them more effectively and with confidence, as AI is no longer a magical black box that just works.

The emergence of machine learning

As we explore the fascinating world of AI, we naturally come to focus on machine learning, the foremost fundamental concept in our ongoing exploration. Machine learning is a subfield of AI. It involves the development of algorithms and models that teach computer systems to identify patterns in data and use this knowledge to make informed decisions or predictions.

Let’s illustrate the concept of machine learning with a simple example.

Imagine you give your friend Alex a bunch of pictures of different animals, cats, dogs, and rabbits, and you tell him which animal is which for each picture. As Alex examines the images, he begins to recognize patterns: cats have pointy ears and whiskers, dogs have floppy ears and snouts, and rabbits have long ears and fluffy tails.

Through repeated observation, Alex becomes pretty good at guessing which animal is which just by looking at their features. Now, if you show Alex a new picture of an animal he hasn’t seen before, he can apply what he’s learned about these patterns to make an educated guess—it might be a cat because of the pointy ears and whiskers, for example.

This process mirrors machine learning, where algorithms are trained on data to recognize patterns, such as features in animal pictures, and use these patterns to make predictions or decisions about new data that they haven’t seen before. This ability to learn from data and predict outcomes is what makes machine learning such a powerful tool across various fields.

Let’s look at a more technical example that illustrates machine learning, focusing on the relationship between input data and outcomes. Imagine we have a dataset containing information about houses: their sizes in square feet and their corresponding prices. The goal is to build a machine learning model that can predict the price of a house based on its size.

We start by collecting the dataset, such as house sizes and their prices. We prepare this data, ensuring it’s suitable for a machine learning model. We then select an algorithm that can predict continuous values, such as house prices based on size. By feeding the algorithm with the dataset, it learns the relationship between house size, the input, and house price, the output. During training, the model fine-tunes its parameters to reduce the prediction error. Once trained, the model can estimate the price of a new house by applying the patterns it has learned.

For instance, if the model has learned that a house’s price increases by $10,000 for every additional 100 square feet, it can predict that a 1,500-square-foot house might be priced at $150,000 more than a house with 1,400 square feet. This demonstrates how machine learning algorithms can learn from data and make predictions on new, unseen data.

To summarize, in machine learning, the goal is to use an algorithm that can predict outputs from inputs. This algorithm is learned from examples of past input-output pairs. Once learned, this algorithm can then be used to predict what the output would be for new input data that the model hasn’t seen before.

The rise of machine learning has been a driving force in the acceleration of AI. The momentum gained by AI with the advent of machine learning is profound, as it shifted the focus from rule-based programming to data-driven learning. This transition allowed computers to evolve from performing predefined tasks to making predictions and decisions based on data analysis. The importance of machine learning in this context is its role in enabling AI systems to uncover insights and patterns within vast datasets that were previously unattainable with traditional computing methods. This capability has not only propelled AI forward but also opened new possibilities and applications, making AI more versatile and effective across various domains.

The rise of deep learning and neural networks

Now that we have seen how machine learning works in general, you may wonder about real-world scenarios where we encounter all sorts of data, such as images, text, and audio, and other forms of unstructured information. The traditional machine learning models often find these unstructured datasets challenging, as they typically require significant manual effort to process.

This is where deep learning comes in. Deep learning is a subset of machine learning that excels in interpreting and learning from unstructured data. You can think of deep learning as the whiz kid of machine learning that is good at making sense of messy data. It uses layers upon layers of artificial neural networks, like a deep stack of filters. Each layer learns different aspects of the data, allowing the system to process and understand all of that unstructured information. These layers help deep learning models to identify complex patterns and make smart predictions or decisions. “For example, deep learning is what driverless cars use to process images and distinguish pedestrians from other objects on the road or what your smart home devices use to understand your voice commands.” (Microsoft, What Is Deep Learning?, https://azure.microsoft.com/en-us/resources/cloud-computing-dictionary/what-is-deep-learning).

Artificial neural networks are composed of layers of interconnected neurons that the input signal goes through one after another to predict the outcome. For example, in the following diagram, to identify whether it’s a flower in the picture, the image passes through multiple layers of neurons (depicted as circles) to generate a prediction:

Figure 1.5 – Visualizing artificial neural networks

Deep learning algorithms are inspired by the structure and function of the human brain, and they are capable of automatically learning patterns and representations from data. These deep neural networks learn from vast amounts of diverse data, enabling them to automatically extract increasingly abstract features from raw data, making them highly effective for tasks such as image and speech recognition and NLP.

Deep learning enables machines to process and recognize patterns in data in ways that are inspired by human cognition, but at a much larger scale and scope. The advent of artificial neural networks has enhanced AI technology considerably, allowing machines to tackle previously insurmountable tasks, thus revolutionizing our use of technology in processing and engaging with intricate data.

Introducing generative AI

Generative AI is a fascinating branch of AI that specializes in creating new content. Leveraging deep learning techniques, it extracts patterns from training data to create new instances that mimic the original dataset. This innovative technology excels at producing diverse outputs that can be text, images, music, and so on, based on the patterns it has learned from the training data. Generative AI has attracted significant attention for its potential to produce realistic and creative outputs, revolutionizing industries such as art, entertainment, design, and scientific research.

At the core of generative AI are several key technologies and methodologies. Neural networks serve as the backbone of generative AI. These computational models are inspired by the complex network of neurons in the human brain. In the context of generative AI, these networks use their learned knowledge to create new content. Imagine them as skilled apprentices who, after analyzing countless paintings, can create entirely new artworks that blend elements from various styles they’ve observed.

The Transformer models, including OpenAI’s Generative Pre-trained Transformer (GPT) series, have become a cornerstone of modern generative AI. The Transformer model is like a super-smart assistant that excels at understanding and generating content. Imagine you’re asking this assistant a question. It carefully analyzes each word and its context in your question to figure out what you’re asking for, then provides an answer.

For instance, if you ask, “Which state is Seattle in?” the Transformer model analyzes the question by focusing on important words such as state and Seattle. It understands that you’re asking about Seattle’s location. Using its extensive knowledge of language and information from vast amounts of text, it finds the most likely answer. In this case, based on its training, it knows Seattle is in the state of Washington and generates an answer, “Seattle is in Washington state.”

The Transformer model, as its name implies, takes input data and generates contextually relevant responses. It is skilled at processing language by understanding the relationships between words and generating responses that are contextually relevant and informative. It can generate clear and relevant text, such as writing articles, answering questions, or composing poetry. The model can condense long pieces of articles into concise summaries while retaining the essential information.

It can understand questions posed in natural language and provide accurate answers based on the context provided. Beyond translation, the Transformer model can comprehend and interpret the nuances of language, aiding in tasks such as sentiment analysis and classification.

Besides Transformer models, there are other popular techniques for implementing generative AI, enabling its wide-ranging applications in fields such as NLP, image and video generation, music composition, and beyond.

In image and video generation, generative AI creates art, realistic images, and videos for advertising and media production. In music and audio generation, it composes music across genres, synthesizes sound effects, and generates natural-sounding voices for virtual assistants and audiobooks. In text generation, generative AI aids in creative writing by producing poems, stories, and articles, while enhancing chatbots and virtual assistants for more natural interactions. In scientific research, generative AI accelerates drug discovery by predicting molecular structures and designing new materials with specific properties. “The use of artificial intelligence (AI) is proving to be a game-changer when it comes to medicine with the technology now helping scientists to unlock the first new antibiotics in 60 years” (Oceane Duboust (2023) Euronews, Scientists discover the first new antibiotics in over 60 years using AI https://www.euronews.com/health/2023/12/31/scientists-discover-the-first-new-antibiotics-in-over-60-years-using-ai).

LLMs

In earlier sections of this chapter, we covered machine learning, a branch of AI. Machine learning enables machines to make predictions or decisions without being explicitly programmed for each specific task. Then, we explored deep learning, a subset of machine learning, that utilizes multi-layered or “deep” neural networks to analyze complex patterns in data. Deep learning is particularly good at recognizing patterns in unstructured data such as images, sound, and text. We then discussed generative AI, which leverages deep learning techniques to produce original and creative content. It generates outputs that can be text, images, music, and so on, based on the patterns it has learned from the training data.

You may now be curious about the roles of ChatGPT and Microsoft Copilot within the extensive AI ecosystem. To understand this further, it’s important to discuss another key element known as LLMs.

LLMs are a specialized form of generative AI, and they excel in generating text that closely resembles human writing. These models, such as GPT, represent an advanced application of the Transformer model architecture, but scaled up significantly in terms of model size and training data.

These models are trained on extensive datasets drawn from a variety of sources, with each model’s capabilities shaped by the data it has been exposed to, including its training cutoff date. By predicting the next token (which can be a word or part of a word) in a sequence, these models become powerful tools for understanding and generating human language. They can be used for a variety of tasks, such as text completion, translation, summarization, question-answering, and code generation.

LLMs predict the next word in a sentence by considering the words that come before it. This is done by calculating the probability of each possible next word based on the patterns it has learned during training. The model uses these probabilities to generate the most likely next word, which leads to the creation of coherent and contextually appropriate sentences.

In the grand scheme of AI, LLMs represent the cutting-edge of what’s possible with generative AI and deep learning, pushing the boundaries of how machines understand and create human-like text. They are a powerful tool in the AI toolkit, enabling a wide range of applications from automated content creation to sophisticated conversational agents.

ChatGPT and Microsoft Copilot are both technologies that leverage LLMs to assist users in accomplishing tasks more efficiently. While both technologies understand and generate human language, they serve different purposes and operate within different scopes. ChatGPT, part of the OpenAI family, is designed to engage in conversational dialogue and perform a range of language-related tasks, including creative writing, problem-solving, and coding assistance. Microsoft Copilot, on the other hand, is specifically integrated into the Microsoft ecosystem. It aims to streamline workflows and enhance productivity by assisting with tasks across the suite of Microsoft applications, such as summarizing emails, drafting documents, and managing meetings.

At their core, ChatGPT and Microsoft Copilot employ sophisticated AI through LLMs for natural language tasks. These models leverage deep learning and generative AI methods to handle linguistic data in ways that mirror human-like comprehension and innovation.

Understanding natural language processing

You may have noticed that we mentioned NLP a few times in the previous sections. NLP is a core field within AI that focuses on empowering computers to understand, interpret, and generate human language effectively. By ensuring interactions are both meaningful and contextually aware, NLP enhances the way machines engage with and respond to human communication. It is like giving computers the superpower to understand, interpret, and talk to us in our own language. NLP is a key component behind how your phone understands your voice commands, how spam filters keep your inbox clean, and even how virtual assistants such as Siri or Alexa chat with you like a friend.

At its heart, NLP uses clever tricks from machine learning to teach computers how to handle the complexities of human language. Imagine algorithms that can read through mountains of text to figure out whether an email is important or junk, or sift through customer reviews to tell you how people feel about a product. These tools use patterns from tons of examples to make smart decisions about language.

But it doesn’t stop there. Deep learning turbocharges NLP by letting computers learn intricate linguistic features directly from raw text data, diving deep into how language works. Think of it like teaching a computer to spot sarcasm or understand the subtleties in a poem. Neural networks, the brains behind deep learning, learn these things by crunching through tons of sentences, spotting patterns, and building their own understanding of language rules.

Then there’s generative AI, which not only understands what we say but can also create new text that is coherent, contextually appropriate, and often indistinguishable from human writing, with LLMs leading the way. These techniques advance NLP by improving both language generation and understanding.

NLP integrates various methodologies from machine learning, benefits from advancements in deep learning for understanding complex language patterns, utilizes techniques from generative AI for text generation, and pushes the boundaries with state-of-the-art LLMs for advanced natural language understanding and generation tasks.

Putting it all together

With a foundation laid across the core components of AI, spanning machine learning, deep learning, generative AI, LLMs, and NLP, it’s clear we are witnessing a profound transformation. AI has evolved from basic data processing to machines capable of producing original, human-like creativity. Machine learning provides the basic tools for pattern recognition and prediction, which are enhanced by deep learning’s ability to learn from unstructured data. Generative AI then elevates these capabilities, producing original content that mirrors human ingenuity, while LLMs channel this creative force into sophisticated, human-like language applications.

Now that we’ve explored each of these core AI components, let’s take a step back to see the bigger picture, examining how these components interconnect and observing the synergy they share:

• AI is the grand umbrella term that encompasses all forms of computational systems that can perform tasks normally requiring human intelligence. This includes reasoning, learning, perception, problem-solving, and language understanding.
• Machine learning is a subset of AI. It involves the development of algorithms and models that enable computer systems to recognize patterns in data and use them to make informed decisions or predictions.
• Deep learning, a subset of machine learning, advances these principles. It uses neural networks with multiple layers (hence deep) to learn from vast amounts of diverse data, especially unstructured data. These neural networks are structured in a way that mimics the human brain’s neural networks, hence the term neural. Deep learning allows the machine to learn from complex patterns and perform tasks such as image and speech recognition.
• Generative AI leverages deep learning techniques to focus on creating new content. Based on the patterns it has learned from the training data, it generates outputs that can be text, images, music, and so on. Generative AI models can produce original and creative content, and they are particularly useful in fields such as design, art, and content creation.
• LLMs are a specialized type of generative AI that processes and generates human-like text, at a level of sophistication that approaches human understanding and inventiveness. LLMs are built using deep learning techniques, particularly neural networks known as Transformers, which are adept at handling sequential data such as text. LLMs can perform a wide variety of NLP tasks, such as translation, summarization, question-answering, and conversation.

Having introduced AI and some of its core components, let’s dig deeper into the practical applications of AI in the real world. In the next section, we will take a closer look at how Microsoft is leveraging AI to develop innovative solutions, particularly within the Microsoft Copilot family.

The Microsoft Copilot family consists of a variety of solutions that enhance productivity across different services. These include Microsoft 365 Copilot, Dynamics 365 Copilot, Copilot in Power Platform, Security Copilot, and GitHub Copilot. Each member of the Copilot family brings its own unique capabilities to the table.

Microsoft Copilot employs an LLM based on OpenAI’s GPT framework to generate and interpret human-like text. This advanced model allows Copilot to understand and respond to user inputs with precision. Its conversational interface is crafted to be user-friendly and mirrors the interaction style of ChatGPT, ensuring an accessible chat experience for users.

In the next few sections, we’ll dive into a selection of solutions from the Microsoft Copilot family. Keep in mind, though, that this is just a glimpse of Microsoft’s rapidly growing Copilot ecosystem. The goal is to highlight a few flagship products that showcase the diverse applications of AI. It is important to be aware that Microsoft continues to introduce new enterprise Copilot offerings across a variety of platforms and services, while existing solutions are constantly evolving with new features and enhanced capabilities. By the time you’re reading this, the Copilot lineup has likely expanded even further with additional updates. Furthermore, be aware that product names and branding may change over time, as is often the case with Microsoft’s ever-evolving technology landscape.

Microsoft 365 Copilot

Microsoft 365 Copilot acts as a powerful personal assistant right within your Microsoft 365 apps – be it Word, Excel, PowerPoint, Outlook, Teams, and more. This personal assistant is seamlessly integrated into your daily work tools. It can help you to draft documents, summarize emails, create presentations, manage meetings, keep track of tasks, review content, analyze data, and conduct web research to name a few.

Let’s say your inbox is overflowing with emails. You need to get to the gist of them, but reading through each one is time-consuming. Microsoft 365 Copilot can summarize your emails, giving you the key points in a concise format. It’s like having a personal assistant who sifts through your emails and gives you a brief on the important stuff.

Imagine you’re in a Teams meeting with several colleagues discussing a new project. As the discussion unfolds, Microsoft 365 Copilot is there in real time, summarizing key points and suggesting action items. Suppose you attend the meeting late due to a prior engagement. Microsoft 365 Copilot can provide a summary of what you’ve missed, allowing you to catch up quickly and participate effectively. As the meeting draws to a close, Microsoft 365 Copilot helps wrap up by providing a summary of the key points discussed and identifying the next steps, including tasks assigned to specific people.

Microsoft 365 Copilot is also a powerful tool for task and project management. It assists you in managing your tasks by setting reminders for important deadlines, keeping track of your to-do list, and suggesting the most efficient order to complete your tasks.

Additionally, it aids in project management by providing realistic, AI-generated task plan recommendations, which can expedite the project creation process and ensure the timely completion of tasks.

There are some amazing examples and innovative ways to utilize Microsoft 365 Copilot out there. Whether you’re drafting a report, crunching numbers, or brainstorming for your next big project, think of Microsoft 365 Copilot as your work assistant, always prepared to help, inspire, and simplify your workday.

Dynamics 365 Copilot

Dynamics 365 Copilot provides interactive, AI-powered assistance across business functions. It acts as a powerful personal assistant right within your Dynamics 365 apps, much like Microsoft 365 Copilot does within the Microsoft 365 apps. It is designed for professionals in sales, service, marketing, operations, and supply chain management, empowering individuals to focus more on the most rewarding aspects of their work and less time on mundane tasks.

Imagine you’re a sales representative and you’ve just finished a productive meeting with a potential client. You have a lot of follow-up tasks to do, such as sending a personalized thank you email, updating the opportunity record with the latest discussion points, and scheduling a follow-up meeting. Normally, this would take a considerable amount of time, but with Dynamics 365 Copilot, you can handle all these tasks effortlessly. “AI helps write email responses to customers and can even create an email summary of a Teams meeting in Outlook. The meeting summary pulls in details from the seller’s CRM such as product and pricing information, as well as insights from the recorded Teams call” (Charles Lamanna (2023), Microsoft, Introducing Microsoft Dynamics 365 Copilot, the world’s first copilot in both CRM and ERP, that brings next-generation AI to every line of business, https://blogs.microsoft.com/blog/2023/03/06/introducing-microsoft-dynamics-365-copilot/). Lastly, you can instruct Copilot to find a suitable time for the next meeting based on your calendar availability and send a meeting request to the client, all done through natural language commands.

It’s not just about sales. Dynamics 365 Copilot extends its capabilities across a variety of business functions. In customer service, it drafts contextual answers to queries, providing agents with AI-powered expertise. For finance teams, it streamlines tedious accounting tasks. And in marketing, it helps create compelling content that resonates with the audience. Dynamics 365 Copilot, when integrated within Microsoft Supply Chain Center, can proactively monitor external factors such as weather and financial trends that could disrupt supply chains. It uses predictive insights to identify and alert on affected orders, and can draft emails to notify partners, enabling swift action to mitigate potential issues before they arise.

Copilot in Power Platform

Copilot in Power Platform is a suite of AI-powered tools that help you create apps, automate workflows, and analyze data, all with the ease of conversational language. It’s like having a tech-savvy friend who can translate your ideas into reality, without you needing to know the nitty-gritty of coding. “Now, if you can imagine your solution, you can simply describe it in everyday language, and copilot can create it for you via an intuitive and intelligent low code experience” (Charles Lamanna (2023), Microsoft, Power Platform is leading a new era of AI-generated low-code app development, https://www.microsoft.com/en-us/power-platform/blog/2023/03/16/power-platform-is-leading-a-new-era-of-ai-generated-low-code-app-development/).

Let’s say you run a small bakery, and you want to keep track of your daily sales, inventory, and customer feedback. You describe your needs to Copilot, and just like that, it whips up an app tailored for your bakery. This app not only logs sales and tracks inventory but also gathers customer sentiments through a friendly chatbot.

So, whether you’re a seasoned developer or a business owner with no coding experience, Copilot in Power Platform is your ally, turning your ideas into digital solutions. It’s not just about building apps; it’s about empowering you to create, innovate, and solve problems in a way that liberates you from technical constraints such as coding.

GitHub Copilot

GitHub Copilot can be perceived as an AI-based coding assistant, a companion that helps you write better code by suggesting entire lines or blocks of code as you type, guiding you through the syntax and semantics, much like a seasoned navigator guiding a ship through uncharted waters.

Imagine you’re working on a project, and you need to write a function to calculate the Fibonacci sequence, a series of numbers where each number is the sum of the two preceding ones. Instead of spending time trying to remember how to write this function from scratch, you turn to GitHub Copilot. GitHub Copilot, understanding your intent, suggests a complete function that does exactly that. It’s like having a knowledgeable coding partner who not only understands what you want to do but also helps you do it faster.

Now, you might be wondering how GitHub Copilot compares to Copilot in Power Platform, as they both offer AI-based coding assistance. GitHub Copilot and Copilot in Power Platform serve different purposes and operate within distinct environments, catering to different user needs.

GitHub Copilot is an AI pair programmer that helps you write code faster, drawing context from comments and code to suggest individual lines or whole functions instantly. While GitHub Copilot is focused on assisting with coding tasks, Copilot in Power Platform is about enabling users to build apps and workflows without requiring coding skills, or with minimal coding involved. Both tools leverage AI to simplify and expedite their respective tasks, but they cater to different aspects of the software development and app creation processes. GitHub Copilot is for writing code, and Copilot in Power Platform is for building no-code/low-code solutions.

In conclusion, the Microsoft Copilot family demonstrates how AI-driven solutions can enhance productivity across various platforms and services. From Microsoft 365 Copilot and Dynamics 365 Copilot to Copilot in Power Platform and GitHub Copilot, each member brings distinct capabilities to streamline workflows and improve efficiency. In the next section, let’s turn our attention to Security Copilot, the digital security assistant leveraging AI to defend against cyber threats and enabling a rapid response.

In today’s dynamic cybersecurity landscape, a security specialist needs to be highly skilled and knowledgeable across a wide range of areas. For instance, in incident response, the responder must be well versed in various attack techniques to effectively analyze vast amounts of data and logs to uncover subtle indicators of attack. The responder needs to be skilled in scripting languages, including PowerShell, Visual Basic Script, and JavaScript, and even embedded macros, as these tools are frequently leveraged by cyber adversaries. A good understanding of operating systems is crucial too, as the responder must be able to quickly recognize anomalies. For instance, while lsass.exe is a normal Windows system process for managing and verifying user logins and security credentials, having multiple instances running could signal a potential threat. The responder must also be skilled at identifying malicious exploitation of Windows net commands, along with commonly abused tools such as scheduled tasks, PowerShell, or PsExec (https://learn.microsoft.com/en-us/sysinternals/downloads/psexec). Proficiency in registry keys is vital for detecting malicious code hiding in a registry key. This is merely the tip of the iceberg. From phishing scams to network breaches, and fileless attacks to driverless attacks, the breadth of knowledge required is staggering. No single person can master it all.

This is where Security Copilot steps in, to empower and assist security professionals. As an example, Security Copilot can analyze complex commands and scripts, summarize their actions, and provide clear explanations. This capability enables security teams to understand the script functionality without the need for manual reverse engineering. This ability to quickly grasp malicious scripts or commands allows the security teams to react promptly, make informed decisions, and minimize the damage. This capability is particularly beneficial since there are often very few people in the security teams who can quickly reverse engineer complex scripts, especially scripts that are obfuscated or encoded.

In addition to analyzing scripts, Security Copilot can also examine files. During an investigation, security analysts must sift through numerous files, particularly executable and less common ones, to pinpoint any malicious files. Leveraging Security Copilot significantly reduces investigation time and the risk of overlooking malicious files.

On top of this, Security Copilot has extensive knowledge about a wide array of attacks. This includes information about established and emerging attack methods, tactics, techniques, and procedures used by threat actors, Indicators of Compromise (IOCs), exploited vulnerabilities, and contextual threat intelligence data from Microsoft Defender Threat Intelligence. Given the complexity and diversity of attack techniques, responders often face challenges in acquiring the depth of knowledge necessary to effectively identify all indicators of attack. Security Copilot can help support security professionals by providing extensive attack expertise.

As illustrated in the following screenshot, security analysts can turn to Security Copilot to uncover ways attackers might move laterally across an organization. The question posed to Security Copilot and its response are highlighted in the screenshot:

Figure 1.6 – Interacting with Security Copilot

Alternatively, responders can request that Security Copilot analyze device events and look for anomalies. For instance, the following example illustrates Security Copilot’s ability to detect encoded PowerShell commands on a device. The following screenshot displays the question submitted to Security Copilot along with the initial part of its response:

Figure 1.7 – Security Copilot examines device events

The following screenshot presents the remainder of Security Copilot’s response, including its findings and additional comments. Security Copilot notes: The command lines are Base64 encoded and are part of a PowerShell command. This is a common technique used by attackers to obfuscate their actions and evade detection. It’s recommended to further investigate these events for potential security threats.

Figure 1.8 – Security Copilot detects an encoded command line

As an AI assistant, Security Copilot excels across many different areas. Its abilities in security incident investigation are only a small part of what it can do. This book provides an in-depth tour of its comprehensive features, enabling you to fully utilize Microsoft Security Copilot and enhance your cybersecurity measures with sophisticated AI tools. Through practical case studies, you’ll learn how to deploy advanced AI solutions for tackling security challenges and elevating your cyber defense tactics.

This chapter explored the evolution of AI, tracing its development from early machine learning techniques to the advanced field of generative AI. Key concepts such as deep learning, neural networks, and LLMs were introduced, highlighting how these advancements have transformed AI capabilities. It also showcased the integration of AI into various Microsoft solutions to drive the development of innovation and practical applications. Additionally, this chapter introduced Security Copilot, setting the stage for a more detailed exploration of its role and features in the following chapters.

Additionally, this chapter aimed to demystify AI by revealing the core concepts and underlying mechanisms that make it work. By gaining a clearer understanding of how AI operates behind the scenes, you recognize that AI is not a magical black box, but technologies grounded in clear, logical processes. This insight will enhance your confidence in using tools such as Microsoft Security Copilot, enabling you to apply your AI knowledge to use these tools more effectively.

In the next chapter, we will dive into the specifics of Security Copilot, demonstrating how it enhances security across various Microsoft platforms. Building on the foundational AI principles covered in this chapter, you’ll gain a deeper understanding of AI’s role in cybersecurity.

• Microsoft, What is deep learning? – https://azure.microsoft.com/en-us/resources/cloud-computing-dictionary/what-is-deep-learning.
• Oceane Duboust (2023) Euronews, Scientists discover the first new antibiotics in over 60 years using AI – https://www.euronews.com/health/2023/12/31/scientists-discover-the-first-new-antibiotics-in-over-60-years-using-ai.
• Charles Lamanna (2023), Microsoft, Introducing Microsoft Dynamics 365 Copilot, the world’s first copilot in both CRM and ERP, that brings next-generation AI to every line of business – https://blogs.microsoft.com/blog/2023/03/06/introducing-microsoft-dynamics-365-copilot/.
• Charles Lamanna (2023), Microsoft, Power Platform is leading a new era of AI-generated low-code app development – https://www.microsoft.com/en-us/power-platform/blog/2023/03/16/power-platform-is-leading-a-new-era-of-ai-generated-low-code-app-development/.
• PsExec – https://learn.microsoft.com/en-us/sysinternals/downloads/psexec.