[mrtksn]

I don't believe that this is the future of computer programming, all those coding assistants feel like companies trying to make mechanical horses and caravans when the combustion engine was invented.

IMHO they should be inventing cars, planes and trains.

Why? Because they write code using tools made to accommodate people and when you are taking out people from the loop keeping those is useless. It's especially evident when those AI tools import bazillion of libraries that exist only to help humans tot reinventing and spending time solved problems and provide comfort when coding.

The AI programming tools are not like humans, they shouldn't be using tools made for humans and instead solve the tasks directly. IE, if you are making a web UI the AI tool doesn't have to import tons of libraries to center a div and make it pretty. It should be able to directly write the code that centers that and humans probably shouldn't be looking at the code at all.

[dmalik]

When is the last time you tried Cursor? It definitely doesn't import libraries. It can if you prompt it to but you have control.

I find it works great it you prompt it one step at a time. It's still can be iterative but it allows you to tighten up the code as you go.

Yes you can still go yolo mode and get some interesting prototypes if you just need to show someone something fast but if you know what you're doing it just saves time.

[mrtksn]

It has been some time since my last try but importing libraries isn't my core concern.

I still feel more comfortable with the chat interface where I talk to LLM and make it generate the code that I end up putting together in a dumb editor because I'm still writing code for human based analysis that will be interpreted by machine and my claim is that if the code is actually to be written by machine for the consumption of a machine, then the human should be out of the code creation loop completely and fully assume the role of someone who demands stuff and knows when its done right and doesn't bother with the code itself.

[darkerside]

That would be the case of this were actually AI. But LLMs actually do procedurally generate language in a fairly human way, so using human languages makes sense to me.

[mrtksn]

The problem with this is that we can't get rid of the baggage of higher and higher level of libraries and languages that exist to accommodate humans.

I agree that it makes sense to use these currently but IMHO the ultimate programming will be human readable code-free, instead the AI will create a representation of the algorithm we need and will execute around it.

Like having an idea, for example if you need to program a robot vacuum cleaner you should be able to describe how you describe it to behave and the AI will create an algorithm(an idea, like "let's turn when bump into a wall then try again") and constantly tweak it and tend it. We wouldn't be directly looking at the code the AI wrote, instead we can test it and find edge cases that a machine maybe wouldn't predict(i.e. the cat sits on the robot and blocking the sensors).

[darkerside]

What makes the AI context window immune to the same issues that plague us humans? I think they will still benefit from high level languages and libraries. AIs that can use them will be able to manage larger and more complex systems than the ones that only use low level languages.