AI 2023. Meet ChatGPT. - page 96

 
Vitaliy Kuznetsov #:

You slept through the whole thing) GPT4 can do that already. Plus, if you write a negative promt in mijorney, it will understand it too. For example --no getting stuck in textures



No, everything is normal here, there are no artefacts. a man tied to a car is not an artefact, but a dolb****sm)))) we are talking about real artefacts, such as headphones passing through a bear's head or more cleverly hidden, not related to texture glitches.
 
Lilita Bogachkova #:

ChatGPT banned in Italy due to privacy concerns

Here we go.

1. Media scandals.

2. Public uproar and proceedings.

3. Intervention by politicians.

4. discussion and drafting of a regulatory bill.

5. Voting and passage of the law.

6. Announcement of government control measures for AI developments.

7. Subordination of IT giants.

8. State monopoly on AI technologies above the GPT-4 level.
 
Andrey Dik #:

No, everything is normal here, there are no artefacts. the man tied to the car - it's not an artefact, but a dolb****sm))) we are talking about real artefacts, such as the passage of headphones through the head of a bear or more cleverly hidden, not related to texture glitches.

Let him look for errors/faults in his own work.

 
I assume that government restrictions on AI development (if imposed) will drastically slow down AI development, as the unprecedented leap is due to massive human participation in learning. Data from user interactions serve to improve the model. Let's imagine that this process stops. Would lawmakers themselves start training AI?)
 
Andrey Dik #:

the question is not "how to create reason, consciousness, thinking" in a machine, but "how to distinguish the reasonable from the unreasonable, the conscious from the unconscious".

the line between living and dead, reasonable and not, is very thin.

what is "alive", what is "dead"?

When you enter parameters into a trading Expert Advisor, you don't expect it to solve problems on any arbitrary topics. but when you ask a question to the gpt, on any topic, it responds. at first it seems that it just gives ready-made information, but no, it is capable of building logical chains, making conclusions. what this thing does, not every person is capable of. who are we to say that it doesn't have a mind? - Of course, it's still a dumb machine, but it's a dumb machine that you want to thank every time you solve a problem successfully.

...

The anitification of human self-reflection happens all the time. It's a psychological syndrome. The attitude of the masses towards AI and myth-making is particularly striking. People who do not know the technology cannot stop animate the computational mechanism. They don't realise that they see an interactive statistical mirror that gives them their own information on demand.

 
Andrey Dik #:

the question is not "how to create mind, consciousness, thinking" in a machine, but "how to distinguish the reasonable from the unreasonable, the conscious from the unconscious".

The line between alive and dead, rational and not, is very thin.

What is "alive", what is "dead"?

When you enter parameters into a trading Expert Advisor, you don't expect it to solve problems on any arbitrary topics. but when you ask a question to the gpt, on any topic, it responds. at first it seems that it just gives ready-made information, but no, it is capable of building logical chains, making conclusions. what this thing does, not every person is capable of. who are we to say that it doesn't have a mind? - Of course, it's still a dumb machine, but it's a dumb machine that you want to thank every time you solve a problem successfully.

...

Read about the principles of LLM and get rid of ideas about the "reasonableness" of these programmes. You do realise that ALL the information that GPT operates on belongs to a human being? That it is taken from the internet, organised, labelled and placed in a training data set? And "learning" here cannot be considered in the human sense. It's a different kind of "learning" altogether. Don't fall into that trap. It's computing the statistical relationships of parts of the data. Nothing else. Poor people in Kenya, for $2 an hour, "trained" NS on texts with negative contexts (racism, extremism, etc.). I mean, they just tagged the data with negative meaning. You're seeing their mind rather than the AI.

Get rid of mythological thinking and realise how much better it is to live as a sane person).

 

Address to all believers in the intelligence of AI:

If you collect all the texts from the Internet and count the statistical relationships between individual words and paragraphs, you get a statistical database (model), in addition to which you can write algorithms that take a user's text and find in the database another statistically related text returned in the response. Or, generate new text based on the statistical signature of the linked texts found.

Of course, they added logic algorithms. But, it's very simple. Anyone can catch it being illogical if they want to, making it a little more difficult.

That's the whole "AI" thing.

 
Реter Konow #:

Read about the principles of LLM and get rid of ideas about the "reasonableness" of these programmes. You do realise that ALL the information that GPT operates on belongs to a human being? That it is taken from the internet, organised, labelled and placed in a training data set? And "learning" here cannot be considered in the human sense. It's a different kind of "learning" altogether. Don't fall into that trap. It's computing the statistical relationships of parts of the data. Nothing else. Poor people in Kenya, for $2 an hour, "trained" NS on texts with negative contexts (racism, extremism, etc.). I mean, they just tagged the data with negative meaning. You see their mind rather than the AI.

Get rid of mythological thinking and realise how much better it is to live as a sane person).

This text could have been written for you by an AI, if the task was to gently induce the opinion that an AI cannot become intelligent. And who would tell the difference)


Who hasn't gotten into the topic yet:

A search engine with built-in GPT for free. This is roughly what Bing looks like

https://googpt.ai/

 
Vitaliy Kuznetsov #:

This text could have been written for you by an AI, if the task was to gently induce the opinion that an AI cannot become intelligent. And who would tell the difference)

...

1. I am forced to "gently" incline to opinion (sanity) so that the branch is not inundated with pseudoscientific nonsense and fantasies that periodically accumulate. A cold intellectual "shower" benefits everyone. Including me. Otherwise, there's nothing to do here.

2. Of course, the AI could have written my posts (at my request), only this in no way proves its reasonableness and does not negate my theses.

 
Реter Konow #:

2. So you have an idea of another technology to be so confident about not meeting your criteria. I like the boldness of your judgement, even though I am very sceptical. Sorry.

3. So when thinking, memory, attention and perception work together, Consciousness emerges. How simple. I won't argue with that, let's say. But, that's in a human, but in a machine, would it be the same? And how do you realise that? No, I'm not going to plagiarise your ideas, I'm just surprised at their boldness. ) You don't have to answer, the question is rhetorical.

4. Well, one of those who succeeded is Sigmund Freud. I guess you won't argue? Have you read it? Very interesting. Only, how do you programme it?) Again, rhetorical question.

5. You didn't mention an animated algorithm. You said "dead" algorithm, which I twisted into "alive" and then into "animate" because I don't see a way to write a truly alive algorithm.

In my opinion, we can only talk about the degree to which an algorithm can be animated by a human, and all AI development will focus on simulations that support a high degree of animation. No one will create a real "live" one. (And who would want to? Rhetorical question.)

2. Yes, my system of ideas is based on a large amount of knowledge, it so happens that I have been interested in these questions and have thought a lot on my way. They may be wrong, of course, so your scepticism is justified. Especially since it would take a huge amount of both of our time for you to familiarise yourself with all the details. And I've said before that I have no goal of convincing anyone systematically and fully. I participate in the topic to the best of my ability and desire.

3. Not human, but some kind of consciousness is formed. It is not some kind of magic. Especially since modern experiments in zoopsychology have found very strong indications of consciousness in one form or another in birds, for example. Having observed domestic dogs for a long time, communicating with them, in a number of cases I observe surprising signs of consciousness. Earlier people did not think about it very much and such ideas sound unusual, but nothing can be done. The spread of information obeys a logistic curve. Apparently, questions of consciousness are at its beginning.

4. You just don't know 🙂 If you dig deep, you will find out that this sex-obsessed fantasist with his flawed theory of personality, has not come up with anything useful and valuable, except for one idea about the presence of unconsciousness in the psyche (somehow you never thought before that there is something under the surface of the ocean too). And his daughter Anna Freud didn't even develop her father's theory of personality, but dealt with the topic of protective mechanisms of the psyche. He had disciples, Jung, Adler, also famous names. Freud was a lover with Jung, hotly quoted in psychological pops. In general, there was a lot of crap, I won't clutter up a serious thread with it.

Besides, there is still no complete theory of personality (in public space). Although, we are talking about a separate phenomenon of psyche. And there is no complete theory of psyche at all, only a hodgepodge of different disciplines, often quite poorly connected, and also contradicting itself. So do not have illusions about the "science" of psychology (which, by the way, in the strict sense is not a science at all).

5. Living, dead - all these are also conventions. There is no single definition. I used these words in such an understanding that the living is a mobile, active, capable of development, phenomenon, and the dead is passive or reactive. I did not mention anything about "soul", I consider this word superfluous in our topic.

Vitaliy Kuznetsov#:

I propose an interesting reasoning here.

1. Current AI as if has access to all digital information, is able to mix, interpret, combine it, creating new pictures (including new species of animals), create new things, phenomena. But it's all in the digital world. And this AI in the digital world is a digital god.

A human being can do all the same things, just very slowly. We are hostage to our biochemical nature. And man can do it by himself. He sits on the couch all day, stares into the void, and then some idea strikes him and he runs off to create another game. But AI can do something quickly, but only in response to a request from outside. In humans, the request comes from within. Feel the difference. Actually, I'm back to my 4 criteria of consciousness, which the AI we know clearly does not have. But it simulates personalities in a funny way 🙂 🙂 .

Retag Konow#:

The spiritualisation of human reflection of self is happening all the time. Psychological Syndrome. The attitude of the masses towards AI and the myth-making, manifests this especially vividly. People who do not know the technology cannot stop animate the computational mechanism. They don't realise that they see an interactive statistical mirror that gives them their own information on demand.

That's right. This syndrome is called "projection" and belongs to the defence system of the psyche. By the way, do you know what exactly the psyche protects by various mechanisms from the attacks of the world around us? 🙂 Self-esteem (the system of ideas about oneself or also called "ego"). Take a look at the real world, around what most of the human activity revolves....

Maxim Kuznetsov#:

let him look for errors/faults in his own work.

We have already seen how it reacts to its mistakes now: with lies and empty apologies. If it were capable of correcting errors in its responses, it would avoid them immediately.

Vitaliy Kuznetsov#:

This text could have been written for you by AI,if there was a task to gently incline to the opinion that AI cannot become intelligent. And who would distinguish)

I.e. both the task and the opinion are pre-prepared by a human, and the "AI" just assembles the letters in a suitable way into a text. The benefits are pouring in 🙂 🙂