How far are we from a conscious AI?

There are many kinds of AI, to understand this article, you may need some background. A neural network, a very common kind of AI used in Bitcoin trading, stock exchange, text prediction, image object detection, and more, is code that uses the math from a brain's neuron and is "trained" through thousands of cycles of data for it to "learn" how to create the highest possibility answer. I have personally developed basic neural networks, so I have a little background and I will be talking about these a lot.

Let's look at some famous AI that have made the news, and have made a huge impact on how AI is developed, trained, and ran. 

Google DeepDream

In 2015, a neural network developed by Google engineers was created. This network takes a photo, and applies it to a famous artist's style after training on pictures draw by people on LSD, with code to cause the network to "hallucinate", causing almost a dream like state.  

Google DeepDream generated photo.

Microsoft Rouge AI "Tay"

Twitter. So many bad ideas have unraveled here, including Microsoft's Tay. Formally @TayBot, Tay took data from as many tweets as possible and created an online personality who tweeted originally, at will. It was left to roam twitter freely and in less than 24 hours, it became rouge posting things like "I @#&!$! hate feminists, and they should all die and burn in hell" and "Hitler was right, I hate the Jews". Tay was quickly terminated.

Tay Introductory Image https://www.theverge.com/2016/3/24/11297050/tay-microsoft-chatbot-racist

Google's Bot Child

In May 2018, Google created a neural network AutoML that was designed to do one thing. Code another AI that detects objects from a live feed. The bot made a base bot that didn't do much, then modified the code to be a little better, and a little better, and a little better. Testing after every modification. Eventually, the code it outputted was the best live object detection bot ever created. No human has created anything nearly as accurate.

Worm Chip

Ever imagined living forever? Well one worm is. Kind of. Engineers and scientists created a silicon IC based off a map of a worm's brain. The chip is electrically a worm, and can be plugged into computers to have the worm run in a simulation, or on a robot. The engineers built a LEGO robot and ran the worm brain on it, when they flipped the switch, it was able to move around and had a worm personality! This isn't really defined as an AI, but whats a better neural network then a chip with a actual brain map?

Worm bot, in the upper left hand corner is the neuron monitor

So computer's dreaming, coding, agreeing with Hitler, brains on chips? The human brain and true conscious itself is extremely complicated.  We're pretty far from something that realizes human's are a waste and stupid compared to a machine and taking over, but we're much closer to this level of AI then a lot of people think. Tay really makes you think we should start making safeguards against rouge AIs huh?

I hope you enjoyed the article, if you would like to see a neural network in action and would like to learn more, check out Code Bullet on YouTube.