I”d have to try to teach someone in their field how to do so, and recognize the

I”d have to try to teach someone in their field how to do so, and recognize the delta between what they’re doing and what I’m doing. With that understanding I could probably articulate it in general terms.

I’ve been working with neural networks since the seventies, and expert systems since the eighties, and on my current project since the nineties. So I already think about pretty much everything in manifolds. And in that sense I sort of intuit how the models function.

And since I work with them every day, i know how to avoid their tendency to go sideways (which is often).

The primary lesson I’d teach it is that you need to provide it a theoretical foundation of whatever paradigm you’re working with before you can ask questions of any depth or meaning. Writing books (what brad and I do) means we spend a lot of time teaching. it something so that we can ask questions about it.

For your field, you’d need to give either work with it to develop a theory or provide it with a theory before you can ask much of meaning within the context of the theory. In other words you need to provide it with the organization and limits of the context in qhich you want to ask the questions.

Reply addressees: @EdwardKulak