I mostly write about mental health on mindpluckd, but I also know a lot about AI. I’ve written on the subject in the past. But now, I’m going to get really regular with it, and start covering everything that’s required to make it in the field. You can think of this particular post as an introduction.
What are the objectives of AI?
Let’s start by drawing parallels with real intelligence. Here, we get into a bit of cognitive psychology which is basically concerned with how the human mind processes information to yield mental states and observed behaviors. With AI, you replace how does the human mind process information to yield mental states and observed behaviors? with how can machines process information to yield internal function states and observed function outcomes? That’s what AI is all about, at its essence. Creating machines that build internal structures out of information and then use those information structures to make decisions/predictions.
Where does math come in?
Well, math is sort of an umbrella term for scientifically precise information processing. If you say, for example, that “I love chocolates, so you should too”, this reasoning is missing a few aspects like: Why should you and I be the same, or even similar at all? What are the measurements that can attest to similarity between us? You see? The only tool in the world today that allows us to be precise as opposed to vague during information processing is math. When an information model is mathematically rigorous in its structure and operation, every single conclusion drawn from that model becomes as precise as it can get.
The computer/machine metaphor for human cognition
The questions/problems are pretty similar actually. For example:
(1) How is knowledge represented in a structure such that it is accurately and efficiently applied during a human task? In AI, this is basically how should data be structured and classified such that it can lend itself to reasoning techniques?
(2) How do humans model mental states during an observed behavior? In AI, this is how can a machine model internal function states internally as it carries out an observable operation?
(3) How do humans model logical reasoning through their knowledge base? The AI equivalent is how can you make well-reasoned decisions and predictions based on the knowledge base?
(4) How do humans perform evidence based reasoning through their knowledge base? Here, the AI equivalent is how can a machine assign probablities based on data evidence to its decisions and predictions?
You see? There is a very real parallel between actual human cognition and machine cognition. But when it comes to the science of machine cognition, math is an indispensable tool because it’s the only tool we have that can model cognitive information systems in machines.
I hope this was a good enough introduction to the field of AI. In the posts to follow, I will start getting into the nitty gritty of it. But don’t worry. I’ll make it clear. None of it is magic. As always, feel free to sound off in the comments below!