I often set out to learn about Artificial Intelligence. I must have been 12 or 13 when I decided I wanted to build a humanoid I can talk to. It might have something to do with the fact that I was fighting with my brothers everyday. It could also have been intellectual curiosity.
In my mind, this machine would start as software. I'd program it to listen, then to talk. Then I'd add a webcam and teach it vision. I'd slowly teach it to do things until it becomes better than me. While it was practicing intelligence, I'd learn robotics and build a humanoid body to encase it. Then I'd have to decide if it was a male or female. While my brothers were playing their video games, these were the thoughts that kept me up at night.
But when I sat at the computer, I had no idea how to get started. In the early 2000s, there wasn't a lot of material to go from on the web. Plus the little that was there was in English. One article I remember translating through yahoo's babelfish had one idea that stuck with me. It said something like "It's hard to say a person is intelligent if you look at them through a microscope." The building blocks of a living being are simple. They perform small simple tasks that can be programmed in a computer. But zoom out and these small components add up to become intelligent beings.
That's just what you see when you explore neural networks, the idea behind the state of the art AI today. They are small simple components that hardly look intelligent. In fact, they are simple math equations. Yet, put them all together and you have face recognition, machine translation, and self-driving cars.
I didn't build my humanoid, turns out I don't need to talk to computers that badly. But that experience has sparked an interest in computer intelligence that I continue to practice to this very day.