How dangerous is AI?

 Over the weekend I watched The AI Dilemma on Youtube. It's about the risks of the race for AI supremacy (or at least advantage).

There's lots here to be concerned about, and their practical illustrations of the risks are nothing short of shocking: not least Snapchat's AI Friend feature cheerfully supporting the grooming and proposed rape of a 13 year old girl by a much older man. Likewise their claim that 50% of researchers working on AI think that there is at least a 10% chance that AI will cause the extinction of humanity.  Yes, read that again! 

I won't try to summarise their argument here; rather I recommend that you watch the video. But beyond that I am not sure what you (or I) can do.  But sharing this seems like a good start.

And perhaps, just perhaps, the existence of thoughtful critics like this (who are not Luddites, but rather wanting proper thought to precede mass rollout) may account for the fact Open AI's Sora - which looks fantastic, but also very dangerous - is not yet released, as it is undergoing safety checks.

One of the many worrying things that Harris and Raskin talk about in the video, however, is the degree to which the cat is already out of the bag, as it were, with social media. That is to say, that we have had a huge rollout of social media before we were collectively ready for it.  And whilst the benefits have been real and significant, the harms were largely unforeseen and even more significant. 

But as I say, I won't try to summarise their arguments here, merely exhort you to watch and think for yourself.