Chat is a bad UI pattern for development tools – Daniel de Laban

The code presses people accurately. That’s good computers need accuracy. But it also forced people to think like machines.

For decades we have tried it by making programming more human friendly. Higher levels of languages. Visual interfaces. Each step helps, but we still interpret human thoughts to computer instructions.

AI would have changed everything. Finally, simple English can be a programming language – one who knows. No syntax. No rules. Just say what you want.

The first wave of AI coding items scandal this opportunity. They make flashy demos but makes garbage software. People call them “good for prototyping,” which means “don’t use it for anything real.”

Many regret AI models, saying we just need to smarter. It’s wrong. Yes, the better AI can make the best predictions about what you mean. But if you build serious software, you don’t like predictions – even smarts. You want to know exactly what you have built.

The current AI items are pretending to write software such as conversation. It’s not. It’s like writing laws. You use English, but you mean terms, establish the rules, and handle complex interactions between all you say.

Try writing a tax code in Chat Messages. You can’t. Even simple tax codes are more complicated to keep your head. So we use documents – we neglect to organize complexity, specified specified points, and changes to changes in systematic changes. Chat reducing you to memory and hope.

This is the main problem. You cannot build real software inaccurate in your choice. Every successful history program tool shows this fact. AI’s short cheat us in thinking we can only chat our way to work software.

We cannot. You don’t programmed by chatting. You program by writing documents.

If your goal is in a document instead of scattering in a chat log, English can be a true language program:

• You can see your entire system once
• You can explain and improve your goal
• You can track changes to the correct
• Teams can work on the system
• Requirements to be their own quality checks
• Changes begin from clear details

The first company to get it owns the next round of AI development tools. They will build tools for real software instead of toys. They will do everything available now look like primitive experiments.