Generative AI is here, along with the growing hype about software development tools and, most intriguing, code-writing tools. The landscape is changing, but hardly anyone can tell in what way exactly.
While some developers worship AI and its power to develop entire features on a whim, others notice compile-time errors, calls made to nonexistent APIs, and outright delusional implementations! What is going on?
In this talk, we will try not to jump to conclusions but see the world as it is. Besides generating a portion of code, can an AI tool keep maintaining it in the upcoming years? Can it generate unit tests that will not break the same instant someone refactors the underlying implementation? (Was that a white box test? Come on, man!) Can a tool write excellent code, even though it was trained on myriads of lines of mediocre code? (Now, this was a trick question!)
Software development is more than writing down some code. In this talk, we shall try to evaluate the state of affairs as of the current date and to find a use for actual AI tools we do have available, not those coming from fairy tales.
Let it not be like writing the description above: Generated by an AI tool; proofread and improved by another AI tool; then written from scratch by a human.
And we didn't even touch on the process of specifying precise requirements...

This talk is a 100% coding session using C# in VS Code with Copilot and Copilot Chat extensions. Please keep this in mind when promoting the talk or selecting the audience.
This talk is the result of year-long experiments I ran and documented in a series of YouTube videos: https://youtube.com/playlist?list=PLSDYwLgFqaX4Ke4h0bVzgadqOesECqKj2