By: Ryan Kienstra on: October 18, 2014 in: Programming
But maybe that’s a good thing.
Sometimes, you want a computer to be “smart.” Auto-correction is useful on phones.
Neural networks can categorize images. But even these come from complicated instructions.
If you write a complex program, a “dumb” machine should run it. It’ll do exactly what you ask.
When something goes wrong, you can find the problem.
There are programs that try to “guess” what you want. Web browsers have to, because much html is poorly formatted.
Browsers fix the problem, but usually in the wrong way. And finding it can be really hard.
Complex programs have many interactions. If one part doesn’t do exactly what you want, it can affect 100 other parts.
So if a computer had intuition about what you wanted, it could ruin the whole program.
We want computers to do complicated things. But we can only understand so much complexity.
It’s not that computers are smart or dumb. They’re a reflection of us.