From what I can tell (l could be wrong) you generally need to have a large data set for most AI and ML. Plus both Android and iOS devices, while stronger today then yesterday, are best suited for lite libraries/versions of the APIs (TF Lite).
So, what exactly are you using this for?
Is this 'simply' to get a better understanding of your products and your clients wishes/habits?
Personally, and maybe only me, I'm not hearing or seeing too many things about this technology.
On the mobile and small computers (Raspberry) the hardware is limited as to what can be computed without too much lag time.
I do find AI and ML interesting, and I've started learning it as well as Python but I'm wondering if it's uses? If you need larger data sets then you're going to look for free historical data, plenty to find, but what are you going to find out about the data that no one else has already found out?
I know I need to learn more regarding AI and ML, and then maybe that will answer some of these questions.
Anyone have any thoughts on AI and ML uses?
“The strongest of all warriors are these two — Time and Patience.” ― Leo Tolstoy, War and Peace
I think ai and ml is used to find subtle patterns that aren't obvious to humans. You do need a lot of data for it to be useful. A good reason that companies are researching ai using computer games is because there is a ton of complicated data, and they can create it from a virtual environment.
I think one overall goal is to find the most optimal ways of find good practice and removing bad practice. Also how to model real environments so that the ai can learn in a virtual environment. For example, Open AI had used a virtual environment that modeled a hand to run the ai. After making many generations using the virtual environment, they translated it over to a physical robotic hand that was able to do the task it was being trained to do.
The technique that they used to develop the ai to do the hand thing was from their research using a computer game, dota2.
Probably the biggest barrier to ai is having data. Whenever there is any research on technique of developing an ai, it is using something that there is a lot of data on. For example, google was going to use drone video footage or something like that and people were freaking out because they didn't want google to make ai to make a drone ai.