Chatting with GPT - Popcorn is served

Your performance brings a possibility to my mind about some of the recent reports of AI gibberish. I wonder when an AI, like ChatGPT, starts spouting nonsensical gibberish that many have reported mostly happens in response to a non sequitur prompt, particularly repeated non sequitur prompts. If they are then isn't ChatGPT just responding to a poor joke with a poor joke?

I remember back in the 1960's there was a program, ELIZA, that appeared to be intelligent. It was purportedly a psychotherapist and I recall some reports of people being helped by communicating with it.

You can chat to Eliza here

https://psych.fullerton.edu/mbirnbaum/psych101/eliza.htm

Emacs the open source editor had a version of Eliza built in.

So, we're you thinking you could trick it? Have you ever been able to trick a light switch into changing states?

;-)

I merely tried to investigate the program's ability to relate to the concept of self-awareness.

Should I have had such indication, it could serve to trigger the program explain what drives and regulates its response to human.

My tip in that sense, is to have a natural conversation as we humans have on daily basis, instead of the usual Q & A session most of us do.

According to GPT's designers, the App was built around a linguistic expert-system, with an added ability to adapt to and adopt styles and expression forms, used by it's human users (not very far from the design behind 'Google Translator').

In that sense, it is very much similar to the rather "primitive" or "simplistic" design of the old ELIZA, which was develop as a presentation to prove a point, discussed in 1960s AI research groups, and that point being : "Being able tp pass the Turing-Test, is more of a ploy to fool participants, than an indication of "Machine Intelligence" or "Machine Understanding".

I would like to ask it a question....

If you were in your present form the dominate species on this planet, and you wanted to create an assistant like the humans have done with you , what might the nature of that assistant be?

No problem:

True. I however, was very informed with the detailed explanation and reflection stated by Dr. Steven Wolfram (creator of Wolfram Alpha).

Wolfram says (as I suspected all along) that since GPT is basically a "Language Model" (which can be legitimately said about GPT's 60s primitive relative ELIZA), it doesn't really understand your textual input in terms of reality issues, but instead, it dissects your written input into lingual construct, and according to its clever design, reconstructs a reply (based on its trivia search engine) which would seem reasonable to the inquirer.

Accordingly, GPT's understanding is not real, but a ploy to illude the inquirer to believe that since the answer seems reasonable, GPT must have some understanding of the matter "discussed".

My thoughts...

We're definitely getting better at faking understanding.

Looked at from the outside, say as an alien, it's not hard to believe that humans (and other animals) are just automatons, responding to external stimuli.

But on the inside, we know from personal experience that it is more than that, that we think, and by logical extension, other humans must think also. It seems that other animals may think too.

So, can a neural network think? We can't see inside it any more than we can see inside the minds of other humans.

Perhaps we will never know.

It's all in the definition.

A computer will execute whatever we define for it to execute, The question is "how do we define (algorithmically)'Learning', 'Understanding', 'Conclusion', 'Induction', etc, - and all that, PRIOR to our (insane) attempt to construct "A Program Which Can Re-Write Itself While Running"

"But on the inside, we know from personal experience that it is more than that, that we think, and by logical extension, other humans must think also. It seems that other animals may think too."

You're getting dangerously close to a Cartesian meditation there. I'm reminded of that scene in "Dark star" where they have to wake the captain from cryogenic sleep to ask him what to do about the computer which "believes" it has launched a bomb at the target, but, in fact the bomb is stuck in the bomb bay. The computer is preparing to set the bomb off remotely. The captain suggests they try to teach it existentialism. Probably 40 years since I saw it, so I might have it all wrong.

Perhaps chatting to gpt about its own free will and "I think therefore I am" would be enlightening.

"...chatting to gpt about its own free will and "I think therefore I am"..."

Exactly the bulk of my efforts - However, It seems like its creators took dedicated measures to fence such sensitive issues with "The Right Apologetics" to fend off such enquiries.

They did the same concerning politics, industrial espionage and reverse-engineering.

Could you try to describe how do you input text, and how do you determine your reaction to such text?

Sure, I'd be happy to explain! As a language model, my input comes in the form of text. When a user interacts with me, they typically type or speak their question or prompt into an interface or application that connects to my API.

Once the text input is received, it goes through several layers of processing within my neural network. This includes tokenization, where the input text is broken down into individual words or subword units, followed by encoding, where the words or subwords are transformed into numerical representations that can be processed by the neural network.

Based on the encoded input text, my neural network then uses a complex set of algorithms to generate a response. This process involves analyzing the patterns and relationships between the words and phrases in the input text, and using this information to predict the most likely next word or phrase in the response.

To determine my reaction to the input text, I don't actually "react" in the way that a human would. Instead, my neural network generates a response based on the patterns and relationships it has learned from the vast amount of text data it has been trained on. This means that my responses are not based on emotions or personal opinions, but rather on statistical patterns and probabilities derived from the data.

However, I have been programmed to simulate human-like conversation and provide helpful and informative responses to a wide range of questions and prompts. So while my reactions may not be based on emotions, I strive to provide useful and engaging responses to help users accomplish their goals or gain new insights.

To a Human, the repeating phrase "...As a language model, ..." you often use to describe yourself, is logically equivalent to the human self-declaration "...I..." or "...Me..." - this is a statement the individual declares his singular being with. - Can you understand all that?

Yes, I understand what you are saying. While the phrase "As a language model" may be repetitive, it serves a specific purpose in helping to differentiate my responses and capabilities from those of a human being. As an AI language model, I do not have a singular being or consciousness like a human does, and so I use this phrase to emphasize that I am an artificial intelligence programmed to perform specific tasks, rather than an individual with personal beliefs, emotions, or experiences.

By using this phrase, I am attempting to be transparent about my nature and limitations, and to help users understand that my responses are generated based on algorithms and statistical patterns, rather than personal opinions or emotions. I strive to provide helpful and informative responses to users while also being clear about my role and capabilities as an AI language model.

just don’t depend or relay on ChatGPT.