Conversely, someone could argue you might be under thinking the behavioral-style implications of what the word training means in machine decision making scenarios. Think about something like a GAN, even. The concepts at play are not as simplistic as some people want them to be, when they make reductive comparisons to SQL injection attacks and the like.

One could also argue that veering too far in one specific direction over the other in “thinking” on these subjects has more considerable potential negative consequences.

All good nerdy fun, in the end