Copilot to Cover - Why AI Can't Replace Developers but Can Make Life Better

Explore the potential and limitations of AI-augmented coding tools in this 43-minute Devoxx conference talk. Discover how products like GitHub Copilot and AWS CodeWhisperer are transforming Java development by streamlining repetitive tasks. Compare different AI approaches in software development, including GPT-based auto-complete tools, Google's Alphacode, and Diffblue Cover. Learn about the pros and cons of each method and how AI can enhance productivity in areas like unit testing. Gain insights into the future of AI in coding, its impact on development teams, and why human developers remain irreplaceable despite technological advancements.

Introduction
The problem with AI
What youll learn
About A Piper
What is AI augmented coding
What is the state of the art
What is Copilot
Transformers
GPT history
Training on GitHub
Running on the cloud
What is it good for
Bullet blade boilerplate
Foreign lands boilerplate
Code Whisperer
Test Writing
Testing as a Search
Reinforcement Learning
How does it work
What its good for
Demo
Pet Clinic
Coding Competitions
Is AI going to replace developers
The intent of coding competitions
How Google uses Transformers
How Google uses Cover
Winning coding competitions
Summary
Raffle
QA