But whether the AI was trained by enough fishery samples and really understand the causality is still a problem. So, it is not the time to make AI guide on fishery/aquaculture practice.
The problem with AI is the fact that it works practically only with data that are stored in magnetic form. There is a large amount of data stored in forms that are not linked to the Internet (non-public databases, printed media, personal experiences etc.). AI can be a very good helper, but a bad master. In order to select from the data those, that make sense and that we understand, i.e. informations, we should not accept the answers that AI gives us without verification and expansion with resources that AI does not (yet) have access to.