When I was in early school grades, a common question in math classes was "Why do we have to learn this, when we can just use a calculator?"
The best answer was "How do you know when the calculator is wrong?", for example when a miskey was entered. Same goes for AI, except knowing enough to spot mistakes or hallucinations takes a foundation knowledge.
I'd love to know who everyone thinks is going to be responsible for creating new language primitives to train AI tools on if no one learns the fundamentals of programming anymore. Someone has to make the primitives.
You can't effectively use the tools if you can't understand and direct what they should do.
What means to learn to code will shift, it will be a shallower knowledge. But a little bit of coding skill goes a long long way when controlling coding agents.
Learning to read code, and ask important algorithmic and architectural questions about the systems one is building will be key. Instead of raw production, the role will shift to curation and editorializing.
5 comments of 10
[ 0.25 ms ] story [ 19.5 ms ] threadThe best answer was "How do you know when the calculator is wrong?", for example when a miskey was entered. Same goes for AI, except knowing enough to spot mistakes or hallucinations takes a foundation knowledge.
What means to learn to code will shift, it will be a shallower knowledge. But a little bit of coding skill goes a long long way when controlling coding agents.
Learning to read code, and ask important algorithmic and architectural questions about the systems one is building will be key. Instead of raw production, the role will shift to curation and editorializing.