1. Are there any drawbacks to in-browser Ai?

Yes, there are a few. This tech is not yet mature. As a result there are bugs and other frustrations to contend with. One being that the apps may not run in all types of web browsers - in some browsers they freeze.

Also, the model has to download to your local machine. No one likes waiting. Therefore, to ensure a good user experience developers have to use small models that will download relatively quickly. A small model may not be as powerful, in terms of prediction accuracy, as a larger one would.

2. Where can I learn the basics of Ai without having to code or do math?

There's an excellent non-technical course on Coursera, presented by Dr Andrew Ng, called Ai for Everyone. It will help you gain the basic knowledge and mental framework needed to develop an Ai strategy for your company. The course can be taken for free.

3. How can I learn to build Ai powered apps at home?

It's not hard to build these apps. Please refer to the resources section of this Kaggle notebook. I’ve also included some practical implementation tips in this GitHub repo.