When I read something about the dangers of AI I can’t help but fear the writers are missing an important point. Saying that AI is bad is like saying the Internet is bad or guns are bad. In truth none of these items are bad. What is bad, and what we need to focus on, is that there are bad people who can and will use these items to amplify what they can do and thus inflict harm on others.

Back before the Internet was known by most of the world those who supported it advocated for all of the good it can do, but we failed to take in to account how it can be used for harm. What is common among AI, the Internet, and guns, particularly automatic guns, is the scale and speed at which harm can be done.

So, my advice is, focus a bit less on the technology and more on the ways in which people want and might use that technology for harm and then craft policies aimed at constraining the the people who may do that harm. It might be making a nuanced argument, but I think it is an important one when encountering those who see themselves as needing to defend/protect the items.