In order to develop Artificial Intelligence learning must be defined. If we think of learning as a natural intuitive thing then it would be mind numbing to try and replicate, rather learning should be created and it will take off by its own. Essentially when building an AI system we are playing nature we are filling in that blank spot. Animals come ready knowing what to do on a basic level, they have the tools they need to learn and for an AI those tools are asking questions and the ability to categorize them.
Learning is in three parts as I currently see; data collection, testing, and categorization.
Data collection is just collecting everything and anything that can be quantified, and everything can be quantified, in my opinion even emotions can be quantified because at the end of the day, in an oversimplified explanation, certain chemicals in our brain set off electrical signals that cause us to act or feel a certain way. Bare in mind this blog post is completely theoretical and is just me freeing my mind, definitely do not take any of this as fact and please do the research. Essentially the more data collected means there is more things to learn from and in turn make the “right” decisions based upon the collected and processed data.
I have thought for a long time about how learning actually takes place and currently I have fell upon the thought that learning can only come from testing or making mistakes and figuring out statistically what is the best decision to make based upon prior experiences or tests. Ultimately it needs to be said that the two basic elements in a test are yes and no. Does the test yield positive or negative results?
The last part is categorization. How could you ever make a decision if you didn’t have a previous test to base it on? If that previous test was not categorized you could never find it, and sometimes you have a situation where the exact same thing never happened before so you can take a very random guess on what to do, or base it off the next closest experience, or a combination of both.
Destination Limbo Signing Off,