Imagine the liability concerns. Only takes a few false positives...
Probably much smarter is a general ML based app that allows user-defined, crowdsourced datasets. So for example, you download the app, and you pay to train a model using datasets that other people gathered. This could be used for things such as middle school science experiments (what is ML? what are the differences between cloud and rock types?) to things like magic mushroom identification. Since the models and datasets are user-curated, you're less exposed to liability, since you merely provide the platform
Probably much smarter is a general ML based app that allows user-defined, crowdsourced datasets. So for example, you download the app, and you pay to train a model using datasets that other people gathered. This could be used for things such as middle school science experiments (what is ML? what are the differences between cloud and rock types?) to things like magic mushroom identification. Since the models and datasets are user-curated, you're less exposed to liability, since you merely provide the platform