You're reviewing _something_ at the end of the day. (Unless you are somehow magically one-shotting problems) If you are reviewing the description of the work, or reviewing the functionality of the app.
I know it's not the same as code, but it is similarly exhausting
Well, the net growth is taken by the AI/tool providers, who indeed bring productivity growth, so the society as a whole see improvements. I guess this is not bad?
Interesting, recently I am working on my own clone of Pac-Man. LLM implementations lose lots of details. They are not 1:1 replication of the original game. For example, the behavior of the ghost is not the same as the original. If I have not implemented the game myself, I can not tell the differences. What LLM produced look like the original game, but they are not.
But it takes -- what, two or three sentences? -- to explain how the ghosts should move. The interesting (and important) thing is not that the model gets it wrong at first, it's how easy it is to correct it.
One-shotting something like Pac-Man doesn't prove much. At the end of the day, one-shot fidelity is going to scale more or less linearly with model size/world knowledge. Why wouldn't it?
It feels like understanding is no longer required to produce code, as long as the program can pass the tests. Also, most software rarely causes severe real world damage, so it simply requires less guarantee and quality than other engineering such as buildings, bridges, etc.. With LLMs, code is just like cars, it carries you to your destination, you do not have to understand how it works.
You can't call people lazy because they use LLMs, just like you can't call people lazy because they travel by train/plane/cars instead of their own feet.
Yeah, it's just how you see programming in your perspective.
It's like lifting weight. Obviously, machines lift weight much better than human for a long time. But people still trains for strength or aesthetics. Even if machine can lift much larger weight and more efficiently, some people still care about how strong they are.
A machine can lift much more weight obviously is not the excuse for them to stop training.
Programming is just the weight lifting for your brain/mind. Even if LLMs can solve all coding problems perfectly, there will be people who still want to program all by hands, for training their mind, or simply enjoying the process. And there will be people care about how good you are at solving a problem with programming.
LLMs being able to code does not mean you no longer need the ability to code.
Cars do not mean you no longer need the ability to walk.
I don't exactly disagree, but I don't think your analogies work.
Machines can lift more weight than humans, but we all encounter situations where strength is useful and machines aren't available or wouldn't be suited to the task.
Cars are faster and easier than walking, but they require roads and are not a possible substitute for all of the walking we do.
But "if LLMs can solve all coding problems perfectly", and assuming that's sufficiently true of cheap/free models, there will be no practical need to code by hand; it'll be something we do purely for the joy of it and/or as brain training (but a niche form of brain training, probably practically inferior to other forms that target skills we still need).
reply