What is AI?
- lmohnani3479
- Aug 13, 2024
- 2 min read
An intuitive explanation for the world's fastest growing technology.
Whether in the media, TV shows, or day-to-day discussion, Artificial Intelligence (AI) has been all the craze in recent months, and while the media-defined characterization of “doing things that humans normally do” isn’t technically wrong, there’s a much more sensible way to approach AI, for both developers and non-coders alike.
Artifical Intelligence usually aims to make a prediction about something or create something new- in essence, it appears like the code is doing something that it’s never been taught before. Many computer scientists say that they don’t train their code what to do, but how to do it, making it seem like the code has human-level capabilities.
Simply put, coders throw a lot of data at their code. For example, I’ll throw 10,000 hand-written, individual digits at my code, and I’ll allow it to make some connections. (If I specify that a hand-written 1 is a 1 to my code, that’s supervised learning. If I want my code to be more independent and figure out what a 1 and 2 in the data mean by itself, that would be unsupervised learning).
After the connections have been made (in our scenario, these mathematical “connections” or “patterns” would include recognizing two circles stacked upon each other as a 8, and recognizing a straight, horizontal line on top of a slanted line as a 7), the training process is complete. With these connections the model has quote on quote “learned” (or really just stored in its memory), the model can be applied to new data that’s it’s never exactly seen before.
We can write a long series of numbers on a piece of paper, scan it, and run it through our AI program. If our code is adequate, it will be able to pick up on those numbers. This seems like magic, because no one told the program exactly what those specific numbers were. But this really isn’t magical, it’s not too artifical, and it’s not that intelligent either. It’s simply an intuitive, mathematical extension of what humans taught the AI program to do.
Comentarios