|
|
|
|
|
by prewett
1242 days ago
|
|
I feel like there was so (informal?) theorem that a full-specification of a program is at least as long as the program. Since you have to tell the AI what the specification is, the worst that could happen is programming becomes more like Knuth's "literate programming", or maybe even better, all programs become formally verified. But that will just mean chasing bugs in your specification. It could eliminate implementation bugs, but it's not going to help with figuring out what my problem actually is. It also isn't going to be able to choose among the solutions, so I'll just end up making a bunch of choices. So maybe programmers become like technical managers, but managing an AI bot instead of a person. Which is sort of how it is today, except the compiler is pretty stupid and pedantic. But even something straightforward: "write a hash function whose output evenly mixes the input" or "write a function to convert an RGB image to a dithered B&W image" is harder than it seems. How do I know the dithering algorithm chosen is good for my data? How do I know the hash function actually does what I asked and doesn't have some funnel somewhere? I'll have to write some tests, but I can't get the AI to write the tests, because how do I know the tests will be right? |
|