How to talk so the AI will listen
A User-Friendly Guide to Understanding and Using AI for Text and Image Generation.
Includes many examples and step-by-step instructions for current AI products such as ChatGPT and DALL-E.
Background
The year 2022 marked a fundamental milestone in the development of Artificial Intelligence. With the release of so-called foundation models and products such as ChatGPT and DALL•E, the most capable AI models we have ever seen are now accessible by everyone - no matter if you are a computer expert or a non-technical end user.
While these systems now can be used with natural language, taking full advantage of the systems nevertheless requires to know how to talk to the AI so that it produces the desired output.
This work is meant to be a compact, user-friendly guide to understand what these new AI systems can do and how you can use them yourself. And even if you do not intend to use them yourself, it will be beneficial to be aware of their capabilities and weaknesses, because — if you like it or not — you will be confronted with their outputs by many other people — your colleagues, your clients, your students, and many anonymous or non-anonymous authors on the internet.
Content
The guide covers the two currently most important areas: Generating text (with many examples for ChatGPT) and generating images (with many examples for DALL•E).
Some of the things you will learn with regard to text are: Summarizing articles, classifying text fragments into categories, turning badly written mails into perfect ones, extracting structured information from free-form messages, getting answers to almost any question, brainstorming with the AI, training how to talk to a certain kind of person, automating research for a specific topic, and generating whole articles from just a few instructions.
Some of the things you will learn with regard to images are: Generating images with different contents and in different styles, creating variations of existing images, changing parts of an image any way you want, and creating large, coherent images in almost any form and aspect ratio.