An overview of Azure AI - Azure Tutorial

- [Instructor] Artificial intelligence has become one of the most transformative technologies of our time, with the potential to revolutionize the way we create content and work. At its core, AI is about creating intelligent machines and software that can perform tasks that would normally require human intelligence, such as recognizing patterns, making decisions, and learning from experience. AI technology is changing the way we interact with the world around us. Whether it's from machine learning algorithms that can analyze vast amounts of data or auto caption services that can parse video and extract captions, we are seeing the massive influence of AI on our lives. What do you think of when you hear AI mentioned in conversation? It's a wide area. My guess is that it's one of these areas: machine learning, cognitive services, large language models, generative AI, or the Internet of Things. Let's start with the difference between artificial intelligence and machine learning. Artificial intelligence is the broader concept of creating machines that can perform tasks that typically require human intelligence, such as understanding languages, recognizing images, making decisions, and solving problems. Machine learning is a subset of AI that involves training algorithms to learn from data and make predictions or decisions without being explicitly programmed for each task. In other words, machine learning is a way to achieve AI. It uses statistical techniques to enable machines to improve their performance on a task over time as they're exposed to more data. Machine learning is important because it allows software to learn and improve from data. This can lead to more accurate and efficient decision-making. By analyzing patterns and trends in data, machine learning algorithms can identify insights and make predictions faster than humans. For example, say we want to classify music into categories like hip-hop, rock, and dubstep. Rather than establishing rules for categorization, developers input thousands of examples of each music genre into the machine learning algorithm. The algorithm then identifies patterns within the music and utilizes those characteristics to recognize and catalog other songs. Machine learning serves as the fundamental building block of Azure AI, providing the backbone upon which many of its key features and capabilities are built. In machine learning, algorithms are used to analyze data and identify patterns or relationships in that data. An AI algorithm is a set of instructions or rules that a computer program follows to perform a specific task. It is a step-by-step process that takes input data and produces output data based on a set of rules. This collection of algorithms is used to train the AI model. The process of training a machine learning model involves feeding it large amounts of data and allowing it to learn from that data over time. The algorithm used to train the model will adjust the model's parameters based on the data it receives in order to improve its accuracy and performance. Once the model has been trained, it can be used to make predictions or decisions based on new data that it encounters. Azure offers a variety of tools and services that allow developers to build, train, and deploy custom machine learning models. One way to create custom models in Azure is to use Azure Machine Learning Studio, a web-based development environment for building machine learning models. With Azure Machine Learning Studio, developers can drag and drop modules to create a machine learning workflow, and then use the platform to train and deploy their models. Another option is to use the Azure Machine Learning Service in the Azure Portal. This is more time-consuming for developers and data scientists. We'll see more about both of these approaches in the machine learning chapter. You don't need to create your own models, however. That's because Azure offers a variety of pre-built machine learning models and services, which provide developers with ready-to-use APIs for tests, like image recognition. With tools like this, Azure can analyze the contents of images or videos and identify specific content. For the last few years, Microsoft has grouped the AI services into categories like Cognitive Services and Applied AI Services. As I mentioned in another video, Microsoft has officially dropped these category names. The features are still available, just renamed to Azure AI Services. However, it's still common to call the group of services shown here Azure Cognitive Services. They are a suite of prebuilt machine learning models and APIs that enable developers to add intelligent features to their applications without needing to build and train their own models from scratch. These services provide a range of capabilities, including vision, speech, and language. Another subcategory of AI is large language models, or LLMs. They are a type of machine learning model that is specifically designed to process and understand natural language. While other machine learning models may be designed to process structured data, such as numerical or categorical data, large language models are designed to process unstructured data, such as text or speech. This requires different techniques and algorithms for processing and analyzing the data. LLMs are typically built using deep learning techniques, such as neural networks, and are trained on large amounts of text data to learn patterns and relationships in language. Some of the Azure cognitive services are based on large language models. For example, the Language Understanding service, LUIS, is based on a large language model that has been trained on a vast amount of text data to understand natural language queries and commands. LUIS allows developers to build natural language processing capabilities into their applications, enabling users to interact with the application using conversational language. Similarly, the Text Analytics service in Azure cognitive services is based on large language models that have been trained on vast amounts of text data to identify sentiment, key phrases, and other insights in the text. This service can be used to analyze customer feedback, social media posts, and other text data to gain insights into customer behavior and preferences. One of the most popular pre-trained language models or LLMs is offered by OpenAI, an artificial intelligence research laboratory. ChatGPT is a pre-trained language model for conversational AI based on the GPT, which stands for generative pre-trained transformer architecture. GPT is a type of neural network. ChatGPT is designed to generate human-like responses to natural language queries that can be fine-tuned for specific use cases. Microsoft is a significant partner with OpenAI, which means that we can use OpenAI models and services from Azure. AI models can also be paired with generation tools, so the vast amount of information gathered via machine learning can also be used to create new content. Because of this, machine learning is having a significant impact on the creative world. For example, machine learning algorithms can be used to generate music, art, and even literature. They do this by analyzing patterns and styles in existing works and creating new ones based on that analysis. One popular generative tool is DALL-E, a neural network-based image generation system developed by OpenAI. It can generate high-quality images from textual descriptions. This system is trained on a massive data set of images and text, and uses a combination of deep learning techniques to generate images that match the given text description. Because of the Microsoft partnership with OpenAI, we can use DALL-E in our applications. Finally, let's look at the Internet of Things. It refers to a group of connected or network set of physical devices, vehicles, home appliances, and other items that are embedded with sensors, software and connectivity. IoT is rapidly expanding, with billions of devices already connected to the internet and many more expected to come online in the coming years. This fast network of connected devices has the potential to transform industries ranging from healthcare and manufacturing to transportation and agriculture. AI is poised to play a crucial role in unlocking the full potential of the Internet of Things. By applying machine learning algorithms to the massive amounts of data generated by IoT devices, we can gain deeper insights into patterns and trends, identify anomalies and outliers, and make more accurate predictions. Azure offers a wide range of AI services, including cognitive services, machine learning, bot services, IoT, and OpenAI. Overall, Azure provides a comprehensive set of AI services that can be used to build intelligent applications across a wide range of industries and use cases.