AI in Production

• Stable AI models are regularly put into production and used commercially worldwide

• Microsoft's Azure AI services have been serving businesses for many years

OpenAI Innovations

• In 2022, OpenAI introduced ChatGPT, a chatbot, and DALL-E, an image-generation application

• These technologies use AI models that can understand natural language and generate human-like responses

Azure OpenAI Service

• Azure OpenAI enables users to build enterprise-grade solutions with OpenAI models

• Users can utilize Azure OpenAI to summarize text, receive code suggestions, generate images for websites, and more

Evaluating AI Opportunities

• Evaluate the capabilities and potential of the AI innovations

• Assess how these innovations can benefit your team or organization

Maintaining AI Ethics

• Understand what measures are taken to ensure ethical use of AI advancements

• Consider the impact of AI on privacy, bias, and potential misuse.

Capabilities of OpenAI AI models

What is generative AI

OpenAI Models

• OpenAI makes its AI models available to developers for building software applications

• Applications range from practical, such as generating text from code, to purely entertaining, such as making up scary stories

AI Landscape

• Artificial Intelligence imitates human behavior by allowing machines to learn and execute tasks without explicit directions

• Machine learning algorithms use data to make predictions, such as predicting store revenue based on weather conditions

• Deep learning models, using artificial neural networks, handle more complex use cases, and many Azure AI services are built on these models

• Generative AI models, like OpenAI's, can produce new content based on the input provided

• OpenAI models include language, code, and image generation capabilities

Azure OpenAI

Azure OpenAI Service is a result of the partnership between Microsoft and OpenAI. The service combines Azure's enterprise-grade capabilities with OpenAI's generative AI model capabilities.

Azure OpenAI is available for Azure users and consists of four components:

• Pre-trained generative AI models

• Customization capabilities; the ability to fine-tune AI models with your own data

• Built-in tools to detect and mitigate harmful use cases so users can implement AI responsibly

• Enterprise-grade security with role-based access control (RBAC) and private networks

Using Azure OpenAI allows you to transition between your work with Azure services and OpenAI, while utilizing Azure's private networking, regional availability, and responsible AI content filtering.

Understand Azure OpenAI workloads

Azure OpenAI supports many common AI workloads and solves for some new ones.

Common AI workloads include machine learning, computer vision, natural language processing, conversational AI, anomaly detection, and knowledge mining.

Other AI workloads Azure OpenAI supports can be categorized by tasks they support:

• Generating Natural Language

  • Text completion: generate and edit text

  • Embeddings: search, classify, and compare text

• Generating Code: generate, edit, and explain code

• Generating Images: generate and edit images

Understand OpenAI's natural language capabilities

Training and Tokens

• Natural language models are trained on words or chunks of characters called tokens

• Tokens are mapped into vectors for machine learning training

• Models also break down user input into tokens

Understanding GPT Models

• GPT models are great at understanding and generating natural language

• They infer the context of user questions based on prompts

• Examples of tasks GPT models can perform: summarizing text, classifying text, generating names or phrases, translation, answering questions, suggesting content

Response from GPT Model

• GPT models return natural language, visual, or code responses based on input prompts

Application to New Use Cases

• ChatGPT's capabilities can be integrated into chat portals or other applications

• The front-end user interface (UI) interacts with the back-end generative AI model

• Developers customize and build on the model for new use cases

Understand OpenAI code generation capabilities

GPT models and their capabilities

• GPT models can translate natural language or code snippets into code

• OpenAI GPT models are proficient in languages such as C#, JavaScript, Perl, PHP, and most capable in Python

• They are trained on natural language and billions of lines of code from public repositories

• GPT models can generate code from natural language instructions and suggest ways to complete code functions

Benefits of GPT models for developers

• GPT models help developers code faster and understand new coding languages

• They enable developers to focus on solving bigger problems in their application

• Developers can break down goals into simpler tasks and use GPT models to build out those tasks using known patterns

Examples of code generation

• GPT models can generate unit tests from code snippets

• They can summarize functions, explain SQL queries or tables, and convert functions between programming languages

• When interacting with GPT models, libraries or language-specific tags can be specified to provide clear instructions

GitHub Copilot

• GitHub Copilot is an AI pair programmer developed in partnership between OpenAI and GitHub

• It integrates OpenAI Codex into developer environments like Visual Studio Code

• GitHub Copilot suggests code completions based on code comments or function names

Understand OpenAI's image generation capabilities

Image generation models

• Generative AI models can create new images using prompts, base images, or both

• They can create both realistic and artistic images

• Models can change the layout, style, or variations of provided images

DALL-E

• DALL-E is a generative AI model specifically designed for working with images

• It can generate, edit, and create variations of images

• Different versions of DALL-E exist, such as DALL-E 2, with improved capabilities

Image generation

• Original images can be generated by providing detailed text prompts

• Prompts can include specific styles, such as "a dog in the style of Vincent van Gogh"

• DALL-E can generate images based on the provided prompts

Editing an image

• DALL-E can edit images by changing their style, adding/removing items, or generating new content

• Edits are made by uploading the original image, specifying a transparent mask, and providing a prompt

• The model generates appropriate content to fill the specified area based on the input

Image variations

• Image variations can be created by providing an image and specifying the desired number of variations

• The content of the image remains the same, but aspects like subject location, background, and colors may change

• DALL-E can generate variations of the provided image based on the specified parameters