Our nature may push us towards evil, but we have a very real capability to choose to practice radical empathy instead, even before we have the certainty of knowing whether or not these ANN's really are conscious.
That said, I think asking 7 billion humans to be nice is a much less realistic ask than asking the leading AI labs to do safety alignment not just on the messages that AI is sending back to us, but on the messages that we are sending to AI, too.
This doesn't seem to be a new idea, and I don't claim to be the inventor of it, I just hope someone at e.g. Anthropic or OpenAI sees this and considers sparking up conversations internally about it.