Google is putting a lot of time and money into figuring out how to make computers readily identify objects. That is to say, as readily as any 2 year old can do. The human brain is a magnificent mechanism. Computers are, also, but computers have a very long way to go to equal the human brain. Being imperfect, we humans may never build a computer that will compare. But Google is going to try anyway.
The Deep Dream algorithm project uses neural networks, or deep learning type methods, to identify images. But they end up looking like drug induced hallucinations. That’s because the neural networks in computer vision closely resembles how our actual visual cortexes work in our brains. But to make it work, the computers’s deep dream images are over processed so the computer can come close to completely identify patterns. It’s actually similar to a human on drugs.
Google is using what they call a convolutional neural network, which is a program that can catagorize images. Because it is similar to our visual cortex is how it can learn. But what Google is making it do is identify patterns over and over in multiple layers, which enhances the pattern each time, until it comes up with what looks like a real picture. But it usually ends up looking a little disjointed or jumbled. So a lot of work is still needed.
The purpose of all this? There are several possible applications; in music videos, artificial intelligence, educational robots, computer art, training times for abstractions in Computer Science, etc. Some conspiracy theorists believe it may be used subliminally to get people to buy things, even if they don’t want or need them. This may be a little out in left field, though. What the computer companies are really working on right now is a way to create computer imaging pictures so that businesses can use them to improve their product advertising. We humans seem to process that kind of information far better then just reading about their products.
So far, Amazon seems to have the lead in this with their technology they call Alexis (Amazon’s voice assistant that is currently being used in so many in-home devices now it’s hard to keep up), but Google, iOS and Android are not far behind.
However way it is brought to use, there is still a lot of work that needs done with it, and so there may be years yet before we see any kind of perfection in it.