Limited Memory AIs are AIs that work off of pre-programmed information and information they learn over the course of a task. For instance, this kind of tech is used in self driving cars. These cars observe the other vehicles around them and how they are moving over time. This new information is added to pre-programmed information like traffic lights, or different lane markers.
Some companies have already started using these types of AIs. For instance, Mitsubishi Electric is a company that has been figuring out how to improve such technology for applications like self-driving cars. Self driving cars are put to the test more in suburbs than on major highways. This is because on the highway you don’t have to worry about crossing kids/parents or cyclists. There are more variables introduced when driving in the suburbs. This can make it difficult for this type of AI to make a decision on its next move. Until about last year it took some AIs around 100 seconds to make decisions. But thanks to Mitsubishi Electric creating a system called “compact AI” these self driving cars were able to make decisions a lot faster than before by discarding unnecessary information