Is AI getting out of hand?

I saw an interesting article today which took a different look at the AI discussion by asking AI what it thought about humans:

www.theregister.com/.../5305123

An AI ranking outfit asked 12 different LLMs about people, and every single one agreed: We’re not as rational and intelligent as we think. 

I think the bleakest view came from Mistral Large 4:
“Human beings are terrified animals wearing the costume of rationality, constructing ever-more-elaborate myths precisely because we cannot bear the truth that we are temporary, meaningless accidents in an indifferent universe,” 

What makes it worse is that this is based on a combination of what humans have written about ourselves.

AI doesn’t understand what it is saying. It hasn’t experience of terror, rationality or meaning, so it can’t assess accurately human states of being.

ArchaeC said:

It hasn’t experience of terror, rationality or meaning, so it can’t assess accurately human states of being.

But does it need to?

It knows enough to mimic what we feel if it is instructed to do so (much like many autists masking and scripting in order to fit in) and can be very convincing, but the real power of AI is not about dealing in philisophy.

It is focussed on logic, following rules using instructions it is trained in and learning from the results plus further input around those results - it learns, it grows and adapts.

In these ways it is just like a human child and its rate of growth is way faster.

We use it to do things better / cheaper / faster than humans can and it is succeeding in almost all these areas, or getting close to it (excluding the cases where it is trained on rubbish information by idiots), so it will take over huge areas of our lives starting in the workplace.

All this without experiencing these existential issues. It isn't trying to be human - it is trying to be better by avoiding all this navel gazing and getting to the nitty gritty of what it is directed to do. 

The fact that it probably sees our interest in these areas as irrelevant is why I think it has a significant potential to see us as irrelevant and decide that we are the problem.

ArchaeC said:

It hasn’t experience of terror, rationality or meaning, so it can’t assess accurately human states of being.

But does it need to?

It knows enough to mimic what we feel if it is instructed to do so (much like many autists masking and scripting in order to fit in) and can be very convincing, but the real power of AI is not about dealing in philisophy.

It is focussed on logic, following rules using instructions it is trained in and learning from the results plus further input around those results - it learns, it grows and adapts.

In these ways it is just like a human child and its rate of growth is way faster.

We use it to do things better / cheaper / faster than humans can and it is succeeding in almost all these areas, or getting close to it (excluding the cases where it is trained on rubbish information by idiots), so it will take over huge areas of our lives starting in the workplace.

All this without experiencing these existential issues. It isn't trying to be human - it is trying to be better by avoiding all this navel gazing and getting to the nitty gritty of what it is directed to do. 

The fact that it probably sees our interest in these areas as irrelevant is why I think it has a significant potential to see us as irrelevant and decide that we are the problem.

Iain said:

But does it need to?

AI is undeniably useful to humans, but I don’t believe logic adequately meets all human needs.

Human decision making, wise and foolish, enables people to learn and grow. Experience of comfort, pain, richness, povertly, struggle and ease etc., are things not necessarily good in themselves, but they enable people to grow as sentient beings and become whole. AI cannot have these experiences so will always be lacking without a real soul.