3 ms·

I am just a layman (per my user name) so I'm not clear what the dangers are with doing this. I am already aware that these language models can be very racist an

I am just a layman (per my user name) so I'm not clear what the dangers are with doing this. I am already aware that these language models can be very racist and sexist. It also makes sense to me that Stack Overflow would ban AI-generated answers because most models are trained on those answers so it doesn't make sense to reinforce whatever learning is going on there. But I'm more confused about the talk about "singularity" and other "doom and gloom" sentiment.

dj_mc_merlin 4y ago
What people mean by "singularity": they are worried the AI will figure out how to make itself more intelligent. By doing so, it might become intelligent enough to augment itself even more, eventually becoming so intelligent it completely eclipses us. A hyper intelligent being which is not human and to some degree immortal is quite a scary concept. This relies on there being more ways to augment its intelligence which can be discovered by it. If it can't augment it's intelligence or knowledge in a more-than-linear way, nothing happens. HN likes this train of thought a lot since it happens in a lot of SciFi books and therefore is realistic (/s). More pragmatically we should be worried about the possible uses of AI such as mass surveillance, disinformation, warfare etc. rather than SHODAN.