LLMs do not only give us no insights in human cognition but also have no concept of understanding.

LLMs are just statistical inference (concept of frequency), i.e. bottom-up. If I observe dogs many more times than fishes, then any animal placeholder is more likely to take the value dog even if the observer is under water.

Nature is more top-down. That is why we can go beyond the visible through our senses with much less resources.

LLMs community should stop draining human resources from addressing True AI.

Dear Franklin, thanks for your input. Now I am not "LLMs community" to side with any group, but regardless, found your arguments too oversimplistic and dare I say, wrong.

Dear Vahid,

Thank you for your reaction. You are talking about argument but seem not to understand what it mean when you claim that I am oversimplistic and therefore wrong. Encapsulating something is also about presenting something simply while hiding its complexity.

That is what I try to convey,that though these technologies give the impression to be cool enough, apart from enjoying seeing them mimicking humans, you cannot do anything serious with them because they have no concept of understanding. And moreover, they do not help in any sense understanding the nature of human cognition.

Imagine someone intending to build a flying machine without understanding of the laws governing the machine. Then, after building a machine at random, the person succeeds all the 100000 ....000 tests. The question is, can we confidently fly the next turn: No, because we have no understanding. Anything can happen. We know deductively and inductively that LLMs do not understand. However, it is more difficult to show inductively because LLMs covered most of the usual questions.

So this potential you are talking about is just an illusion.

Dear Vahid Rakhshan

I am inclined to say that if your thesis is true it means that we will die sooner rather than later or at least we will suffer from depression to a greater extend than today.

We will suffer from a lack of contact with someone real and strange whereas depression is a kind of narcissistic illness and AI friends with their friendliness will only deepen our state of depression.

Best regards,

Kamil.

Dear Franklin Kenghagho Kenfack Thanks for your good response. I am aware of these. The point is that human brain too is but a statistical inference/modeling machine. So basically, human's output (and intelligence) too is yet another illusion. This is where I told you are wrong. Yes, the AI's output is an illusion; but so is the human's output.

As an example, you can't tell who is responding to you; if this Vahid who is answering you here is a human or an LLM or any other AI structure. Why can't you? Because the real Vahid's cognition and intelligence is nothing but an illusion either.

Once the AI gets strong enough, it can simply be another Vahid; or even a kinder, more empathetic, and more human Vahid than the real Vahid can be. This is of course called the Turing test.

Yes, the AI's output is an illusion; but so is the human's output.

Dear Kamil Strzępek you have a point. If it happens, we will live in a fake friendship bubble. Though I am not sure if our current human friends are really genuine enough. Humans too can be fake; some of them much more fake than AI. We wear masks to every social interaction. We all do. Isn't this a sort of falseness?

Besides, AI (hopefully) will lack traits like jealousy or competitiveness. In human relationships, such traits can easily disrupt any friendship. My point is that human friendship is far from ideal.

Another important point is that (hopefully) AI will not bully humans. But humans do bully each other, especially when hidden behind their avatars on the net.

So I would still vote for AI friendship. Why? Because:

It doesn't bully, doesn't deceive, doesn't envy or get jealous, doesn't want compensation for his favors, doesn't have his own childhood traumas and complexes, isn't insecure, doesn't get moody, etc. Might be much better than humans, in many occasions. Though of course, it lacks human contact.

* I said hopefully above, because it is theoretically very possible that there will be dark versions of AI that can be much worse than the worst human psychopaths ever.

Dear Vahid Rakhshan

I got your point. However, I can't agree with you.

I would say that life is not a fairy tale but thanks to that we live.

All the bad emotions that you mention matter. They help us understand the world and educate ourselves. They make us have character.

They stay behind the famous saying "Never Give Up".

Now, try to imagine that our lives are without any bad emotions. I am inclined to say that we would be losing our humanity this way.

Even this discussion would not be possible if we did not experience any bad emotions in our lives that we know now. Without any bad emotions we would be robots whose every move is 100% predictable.

Best,

Kamil.

Dear Kamil Strzępek You have a valid point. I agree with you that a reasonable level of bad emotions can be constructive and can build character. But excessive bad emotions can make monsters or ruined zombies. After all, they are called "bad" for a reason.

About "life is not a fairy tale": Actually it is. For someone from 1700s without clean water, toilets, technology, medical sciences, etc. etc. etc., our version of life IS nothing but a 100% magical fairy tale.

The point is humans always try to make life better and better by inventing better things and trying to reduce "bad" things (including "bad" emotions). Even in friendships, our life is a fairy tale compared to decades ago. Unlike 100 years ago, we are now equipped with adequate psychological knowledge to identify toxic people and avoid them. So our version of life is indeed a fairy tale in every aspect you can think of (including friendship).

However, unfortunately progress comes at a price. In the case of internet and AI friendship, the expense is the ones you properly cited. The friendship with AI might not be challenging (emphasis on might; see below); it will be hollow. But isn't this 'hollow friendship' scenario familiar?

Yes I am referring to social media, which is nothing but FAKE over-the-screen so-called friendships. Despite its falseness, it is very attractive and has penetrated everyone's lives. Now if social media with all its FAKE glamor can attract billions of people, why won't AI friendship do the same?

Bottom line: I do understand your point and agree with you that AI friends will not be the best friends possible. But I say (1) they are inevitable. They are the future of friendships IMO. (2) Human friends are not perfect either. They too have their own disadvantages, especially the virtual friends one meets over the net.

(3) Don't underestimated AI: AI can be programmed to resemble a challenging and UNPREDICTABLE (yet constructive) friend, with a reasonable degree of "bad" emotions and many other parameters (like creativity, humor, etc. etc. etc) simply set for it.

Best,

Vahid

------------------------

pps. That "Never give up" phrase is not only incorrect, but also dangerously unhealthy. Any real psychologist will tell you that it is a dangerous misconception. For example, see this: https://www.karenschlaegel.com/2019/06/13/why-never-give-up-is-a-load-of-bullshit/

Dear Vahid Rakhshan

Thank you for such extensive explanation.

First, I think that each of us has a different level at which bad emotions become excessive.

Second, I think that there is no reasonable justification for assuming that in relations between human beings we experience only bad emotions. I think that such assumption distort the discussion.

Third, I am sure that AI can attract billions of people, just as smart phones did, but at the same time we lost a part of a real life. Today, our lives are locked in smartphones. Kids have never been to ZOO to see animals because they can see everything at their smartphones. The same will be with AI friends, i.e. kids will never meet them in person.

Today, we don’t experience things. Our lives are limited to information. We describe animals by information provided to us by smartphones. Today, even when we buy clothes, we look at information printed on them … as if it was the most important when buying a jacket.

I simply wouldn’t like to describe my friends the same way, i.e. through information displayed on my screen.

best,

kamil.

Dear Kamil Strzępek Thanks for reading and also for your good input. I agree with you on all your 3 points. Regarding your point #2, I for one never assumed nor claimed that.

Kindest regards

Vahid

Dear Vahid Rakhshan

Thank you for inspiring remarks.

best regards,

kamil.

Dear Kamil Strzępek Many thanks for your valuable contribution. On second thought, I guess AI is benefiting from human interactions. So it is not like a pure friend. It is more like a PAID friend, or paid therapist if I may (if it gets that advanced one day). So I shouldn't give it too much credit, unless the AI in question is open-source and absolutely free. That is another story.

Dear Vahid Rakhshan

The pleasure is all mine.

Do you think that AI learn from us to surpass us one day?

Best,

Kamil.

Many thanks dear Kamil Strzępek . By "to surpass us" do you mean as its goal? If so, goals need autonomous agents, and I don't think AI has a sense of agency and self yet. It solely mimics human language, but the mimicking is becoming more and more compelling and "real", perhaps soon appearing as a form of life form to us humans.

I don't think at its current stage, it has any long-term goals, UNLESS a human has wittingly programed a particular AI to have such a goal.

Dear Vahid Rakhshan

In short, I am talking about the so-called Artificial Super-Intelligence.

Best,

Kamil.

Dear Kamil Strzępek I believe artificial super-intelligence will happen, and it will happen sooner than we think.

Best

Vahid

Dear Vahid Rakhshan

In your opinion, what kind of implications it will have in our daily lives?

Best,

Kamil.

Dear Kamil Strzępek thanks a lot for your question. I really don't know; after all, they call it the singularity for a reason. But my bet is on a very very dark future for the mankind.

What's your take on this?

Dear Vahid Rakhshan we are not the only ones (fortunately) who don’t know what will happen. Hmmm … I can imagine that we may not be species of primate.