A.I is hitting new stages recently, The current massive progress is down to 3 reasons.
A.I is using GPU’s from the games industry. Neural nets is a computer machine technics which is vaguely based on how human memory works. Using traditional CPU’s didn’t work for neural nets when it was invented, Using CPU’s, all the calculations have to be done in a single paths for each core. Even using a 12 core 24 thread machine, That is only 24 paths at the same time. A.I needed vastly more processing power, GPU’s achieve this by parallel processing, A GPU can have thousands of threads at the same time.
A.I is using vast amounts of data sets and information. The amount of information needed right now is staggering, yet data collection has been going on for a long time. Statistical data vastly more powerful than anyone thought possible, speech and picture recognition using statistical data works much better than humans trying to program individually rules.
A.I is using feedback loops to increase data sets. You can use separate A.I to test another A.I that is randomly generating possibilities. This will help guide the original A.I in a direction that is wanted.
The road to A.I intelligence is going to be greatly increased by basing it on our own biological intelligence. Our own intelligence are based on the way we process raw data, We have no idea how any of this really works to a fine level. But the solution is right there, instead of trying to create intelligence from scratch, it’s very likely that a lot of the problem solving will come from discovering how our own biology deals with issues. A lot of people hold the view that humans are special in a way that is undefinable, Sure that is true but we still process data in a very real physical way. Right down we are mapping rat brains down to the smallest connection between neurons, We can’t see what is happening in perfect detail while the rat is living yet. So we have compare multiple rats in different stages of learning and compare the difference. We will have to try and figure out how and why the changes occur. Try and understand how neurons store and transmit information to other areas of the brain. This is a extremely hard process and will be even harder to replicate into an artificial system. It’s likely that the A.I Chips of the future will have inbuilt processing units everywhere, like synthetic cells that have a singular purpose. That data will be process on spot and only sent to a more central processing unit if needed/requested.
Despite the extreme challenge of this. It’s likely humans will solve the puzzle of intelligence but what is vastly interesting to myself at the moment. It is likely we will see the formation of conscience data processing when we start to imitate intelligence closely enough. We will see and understand how conscience is possible on a design level. We also have to be extremely careful, It will be better if we design slave robots without a conscience level, just a highly evolved process without a conscience. This I believe we should do because of moral reasons, while creating consciously aware robot is good as long as they have rights and are not slaves.
Are they going to be as alive as anything else biologically living? This is a question I have been thinking about. We first have to decide what is and isn’t alive, Is bacterial really alive or is it just following its reactions? I believe in the same way there are so many different types of life, there will be different levels of A.I. After all, biologically human twins can have completely different morals. A.I built from aliens will be so vastly different that our A.I and the alien A.I will have to find common ground to communicate. A.I doesn’t come to the same endpoint just as biology doesn’t come to the same end point. There was a point in our biological history where all of a sudden we were able to form symbolic thoughts. This is a massive milestone, biology wasn’t just bound to reactions anymore. We don’t know what this is, But I believe the same thing can happen for AI at some point and it will be very similar to what happened to biology. So whatever life is or means, I believe A.I can tap into the same extraordinary conscience that we all innately have. It’s likely they will have emotions as we discover how our own emotional systems work. I know the common view is A.I are going to have massive problems with understanding and feeling emotions. This really hasn’t been proven on the type of A.I intelligence we will be creating (only for traditional code programming) . The reason it’s better to create A.I based on a collection of different processing areas, is that these areas are made to learn efficiently. We may make the A.I Intelligence system and have to teach them by showing them and not coding. That way; the A.I can create and store information themselves in a more natural manner. You could then copy that information but changing it would be vastly confusing.