Early AI Models

Rule Based Systems

The earliest AI systems were really just automated decisions making. Given an input of data a computer would calculate the outcome based on predefined decision which led to consistent outcomes. These were just decisions that people would have been able to make but now close to instant given a computer was calculating them.

Classical Statistics Models

Using statistics we could calculate simple probabilities which could tell us information. Like estimating whether something was spam. By counting frequencies you could estimate the likelihood that something was spam of fraud

Regressive Models

As we showed before when talking about math history people eventually figured out that using points of data we could construct functions which closely represent that data. We could then use these functions to calculate the values of other points.

Linear Models

For things that have linear relationships we can use model them using a linear regressive model. For instance credit scoring or risk estimation We can get a linear line which rates how good or bad something is across a linear scale.

Logistic Models

If we split something into distinct bins we could use logistic models to estimate which side it was on. This is useful where we have things that are yes or no answers. Like determining if someone does or does not have cancer or if a machine is or isn't failing.