Man vs Machine: Can ChatGPT write better SwiftUI than you?
summary
Compares ChatGPT and a human on building a SwiftUI analog clock: ChatGPT spits out 48 lines of code that initially don't compile, then iteratively fixes errors after pasting compiler messages; the human writes about 34 lines, uses relative sizing, adaptive colors, and a sweeping second hand, fixes a SwiftUI rotation bug, and adds a performance tweak (lowered TimelineView frequency). Result: machine generates working code after back-and-forth, but human code is shorter, more robust, and scalable.
AI-written summary · about
chapters (10)
- 0:00AI coding assistants intro
- 3:13Testing ChatGPT with SwiftUI clock
- 6:25Debugging ChatGPT’s SwiftUI code
- 9:36Human vs AI SwiftUI code
- 12:07Scalable SwiftUI clock
- 17:47Apple-like design details
- 21:40Hour markers and positioning
- 27:20Fixing SwiftUI bugs
- 30:49AI code limits
- 34:39AI assisting developers
related talks
01minteresting experience with ChatGPT
01h 12mHow ChatGPT can make you a better Software Developer
50mIdentifying and fixing Issues in Code using AI based tools
05mChatGPT is poorly designed. So I fixed it
39mHow ChatGPT helps to boost your development efficiency
25mCan ChatGPT build a Data Platform faster than a developer?
47mChasing the Main Thread - Adventures in AI Assisted Coding
01h 01mC++hatGPT: How AI Tools Impact Your Life as a Systems Programmer