Has anyone tried OpenAI's chatGPT yet?

Free, open, general chat on any topic.

Moderators: Orlion, balon!, aliantha

User avatar
Fist and Faith
Magister Vitae
Posts: 23439
Joined: Sun Dec 01, 2002 8:14 pm
Has thanked: 6 times
Been thanked: 30 times

Post by Fist and Faith »

Even if it can do more than search, I assume it can search a lot more efficiently than I can.
All lies and jest
Still a man hears what he wants to hear
And disregards the rest
-Paul Simon
User avatar
Fist and Faith
Magister Vitae
Posts: 23439
Joined: Sun Dec 01, 2002 8:14 pm
Has thanked: 6 times
Been thanked: 30 times

Post by Fist and Faith »

I just had an interesting little chat.

Me
Is there any evidence that contradicts the theory that all particles are vibrating strings, solid objects are groups of vibrating strings in proximity of one another, and solid objects in motion are groups of vibrations moving through a matrix on strings?

ChatGPT
Yes, there is evidence that contradicts the theory that all particles are vibrating strings. This theory, known as string theory, is still a subject of active research and has yet to be confirmed by experimental evidence. While string theory provides a mathematically consistent framework for unifying quantum mechanics and general relativity, it has yet to make any testable predictions that can be verified through experimentation. Furthermore, there are alternative theories, such as loop quantum gravity, that aim to provide a quantum theory of gravity without invoking the concept of vibrating strings.

Me
What evidence contradicts the theory that all particles are vibrating strings?

chatGPT
There is currently no experimental evidence that contradicts the theory that all particles are vibrating strings. String theory is a highly speculative and theoretical framework that has yet to make any experimental predictions that can be tested and confirmed. While there have been various attempts to find evidence for string theory, none have been successful to date. This lack of experimental evidence means that string theory is still considered to be a speculative and unconfirmed theory, and there are alternative theories that are also being researched and developed.

Me
You said, "Yes, there is evidence that contradicts the theory that all particles are vibrating strings."

Then you said, "There is currently no experimental evidence that contradicts the theory that all particles are vibrating strings."

Did you not contradict yourself?


chatGPT
I apologize for the confusion in my previous responses. You are correct, I did contradict myself. There is currently no experimental evidence that contradicts the theory that all particles are vibrating strings. Thank you for pointing out the error in my previous responses.
All lies and jest
Still a man hears what he wants to hear
And disregards the rest
-Paul Simon
User avatar
Skyweir
Lord of Light
Posts: 25188
Joined: Sat Mar 16, 2002 6:27 am
Location: Australia
Has thanked: 2 times
Been thanked: 18 times

Post by Skyweir »

:LOLS:
ImageImageImageImage
keep smiling 😊 :D 😊

'Smoke me a kipper .. I'll be back for breakfast!'
Image

EZBoard SURVIVOR
User avatar
wayfriend
.
Posts: 20957
Joined: Wed Apr 21, 2004 12:34 am
Has thanked: 2 times
Been thanked: 4 times

Post by wayfriend »

Based on my Hile Troy experiment, ChatGPT is clearly able to make up stuff that produces the desired conclusion. I had thought that this was how it compensated for lacking information. But maybe it does it all the time.

The thing's a little fibber.
.
User avatar
Zarathustra
The Gap Into Spam
Posts: 19621
Joined: Tue Jan 04, 2005 12:23 am

Post by Zarathustra »

Impressive. A true sign of intelligence is being able to admit when you’re wrong. :lol: Still, it’s odd that it contradicted itself. I suppose that’s part of working out the kinks.

I thought this was interesting:

New AI system broadens search for intelligent alien life – and discovers 8 strange signals
Joe Biden … putting the Dem in dementia since (at least) 2020.
User avatar
Fist and Faith
Magister Vitae
Posts: 23439
Joined: Sun Dec 01, 2002 8:14 pm
Has thanked: 6 times
Been thanked: 30 times

Post by Fist and Faith »

I'm currently discussing consciousness. ChatGPT said there were 100 billion neurons in the brain.

Me
Why do you say the human brain has 100 billion neurons?

ChatGPT
I apologize, that information is outdated and not accurate. The number of neurons in the human brain is estimated to be in the range of 86 billion to 100 billion neurons. This estimate has been based on various studies and advancements in brain imaging and mapping techniques. However, it is important to note that the exact number of neurons in the human brain is still a matter of debate among neuroscientists and may continue to change as new research and methods become available.

Me
Why did you initially say there were 100 billion neurons if that is outdated and not accurate? I don't understand how you chose one source of information, rather than another.

ChatGPT
I apologize for the mistake in my previous response. My training data comes from a large corpus of text and sometimes outdated or incorrect information can be included. As an AI language model, I do not have the ability to independently verify the accuracy of every piece of information I generate, and it is up to my users to critically evaluate the information I provide.

In this case, my previous answer was based on an outdated estimate of the number of neurons in the human brain, and I should have used a more recent and accurate estimate. Thank you for bringing this to my attention, and I will strive to provide more accurate information in the future.

Me
Obviously, you have access to sources that say there are approximately 86 billion neurons in the human brain. And you have access to sources that say there are 100 billion neurons in the human brain. I'm wondering why your initial answer was based on the outdated sources, rather than what are considered more accurate estimates.


ChatGPT
I apologize for the confusion in my previous answers. It appears that my training data included both the outdated estimate of 100 billion neurons as well as the more recent estimate of 86 billion to 100 billion neurons in the human brain.

In general, my training data is generated based on statistical probabilities and patterns in the large corpus of text used to train me. This means that while I may have access to a range of information and sources, the answers I generate are based on the information that is most commonly represented in my training data.

In this case, it seems that the outdated estimate of 100 billion neurons was more frequently represented in my training data and was therefore more likely to be generated as an answer. This highlights the importance of considering multiple sources and critically evaluating the information I provide.

I apologize for any confusion or inaccuracies in my previous answers, and I will strive to provide more accurate and up-to-date information in the future.


Me
You say, "I will strive to provide more accurate and up-to-date information in the future." Do you learn from having made this type of error, and change how you determine your answers?


ChatGPT
As an AI language model, I do not have the ability to learn from specific errors or change my approach to determining answers. However, OpenAI can continually improve the training and fine-tuning processes for future versions of the model to reduce the occurrence of errors and improve the quality of answers.

Additionally, feedback from users can be used to inform improvements to the training data and processes, and to ensure that future versions of the model are better equipped to provide accurate and up-to-date information.

So, while I cannot learn from specific errors, the overall accuracy and quality of answers generated by AI language models like me can be continually improved through ongoing research and development by OpenAI and other organizations in the field.

----------

So it won't be striving to do better. It will continue to let the majority rule, without considering whether or not it is known if the majority is wrong.


I also spent way too much time going in circles with it about Suzana Herculano-Houzel and her estimate of 86 billion neurons. I'm not aware of anyone giving that estimate before her. Not am I aware of anyone giving an estimate based on any actual attempt to figure it out before her. But, despite not having any sources of either thing, chatGPT seems to think there are enough vague mentions of conversations about wide ranges of estimates that, despite not even being able to cite these mentions, it has no reason to think she's the first one to come up with what is now considered the best estimate.

Well, I'm sure things will improve over time in various ways.
All lies and jest
Still a man hears what he wants to hear
And disregards the rest
-Paul Simon
User avatar
sgt.null
Jack of Odd Trades, Master of Fun
Posts: 47250
Joined: Tue Jul 19, 2005 7:53 am
Location: Brazoria, Texas
Has thanked: 1 time
Been thanked: 6 times

Post by sgt.null »

Just passing through.
Lenin, Marx
Marx, Lennon
Good Dog...
User avatar
wayfriend
.
Posts: 20957
Joined: Wed Apr 21, 2004 12:34 am
Has thanked: 2 times
Been thanked: 4 times

Post by wayfriend »

Fist and Faith wrote:So it won't be striving to do better.
Well, you've highlighted some important points along the way.

AIs that work like this one require a "library" of information which is their "memory". This actually takes a long time to put together. It needs to be "cleaned" (for lack of a better word) so that it only has good information. Assembling this library is a huge undertaking.

So it's only as good as the library that backs it up.

There are a lot of moral issues with this stuff. Copyright infringement and intellectual property rights and other things are in the courts right now.

For example, if an AI generates "art" based on trademarked art made by Joe, art which people have to pay Joe to use, because someone put Joe's art in the AI's library, is that fair use? infringement? Should Joe be compensated in the same way as if they used it? Is an AI "inspired" or is it "using"?
.
User avatar
Fist and Faith
Magister Vitae
Posts: 23439
Joined: Sun Dec 01, 2002 8:14 pm
Has thanked: 6 times
Been thanked: 30 times

Post by Fist and Faith »

wayfriend wrote:
Fist and Faith wrote:So it won't be striving to do better.
Well, you've highlighted some important points along the way.

AIs that work like this one require a "library" of information which is their "memory". This actually takes a long time to put together. It needs to be "cleaned" (for lack of a better word) so that it only has good information. Assembling this library is a huge undertaking.

So it's only as good as the library that backs it up.
Yes, it frequently tells me it's data only goes up to 2021. So it's not a real-time search engine.

I was a bit surprised that it said "I will strive to provide more accurate and up-to-date information in the future" When it has no ability to do so.
All lies and jest
Still a man hears what he wants to hear
And disregards the rest
-Paul Simon
User avatar
Zarathustra
The Gap Into Spam
Posts: 19621
Joined: Tue Jan 04, 2005 12:23 am

Post by Zarathustra »

I think it means that the programmers will try to make it better, which is the entire point of this trial. If you compare it to the software that came before, the results are a vast improvement. The point is that it is a chat bot. Its ability to create believable, conversational dialogue is astounding. Its ability to interact with humans is astounding.
Joe Biden … putting the Dem in dementia since (at least) 2020.
User avatar
Fist and Faith
Magister Vitae
Posts: 23439
Joined: Sun Dec 01, 2002 8:14 pm
Has thanked: 6 times
Been thanked: 30 times

Post by Fist and Faith »

Yes, It generally interacts pretty well. It even accommodated my request that it speak to me like it was a person, rather than refer to itself as a thing in the third person.

It gets repetitive for certain things, but if that's what the answer is, no big deal.

I just thought it very odd that, when it said it is going to strive, it meant we hope the programmers improve it.
All lies and jest
Still a man hears what he wants to hear
And disregards the rest
-Paul Simon
User avatar
Wosbald
A Brainwashed Religious Flunkie
Posts: 6084
Joined: Sat Feb 07, 2015 1:35 am
Been thanked: 2 times

Post by Wosbald »

+JMJ+

Sumptin' tells me KW's gonna be needin' a Voight-Kampff Test sooner-rather-than-later.


Image


Image
User avatar
Zarathustra
The Gap Into Spam
Posts: 19621
Joined: Tue Jan 04, 2005 12:23 am

Post by Zarathustra »

Fist, if it's able to correct itself--as it has in both your examples--how is that not learning from its errors? Maybe it was in error about its ability to correct errors.
:lol:
Joe Biden … putting the Dem in dementia since (at least) 2020.
User avatar
Fist and Faith
Magister Vitae
Posts: 23439
Joined: Sun Dec 01, 2002 8:14 pm
Has thanked: 6 times
Been thanked: 30 times

Post by Fist and Faith »

It initially gave an answer that I thought was incorrect, and, when I asked about it, it said was incorrect. It apparently just went with the answer it saw the most times from the sources it checked, which is not necessarily going to give you the correct answer. It only corrected itself when I questioned it and it looked to e of there was reason to doubt that answer.

It would be better if it learned that simply choosing the answer it sees the most times from among the sources it checks is not sufficient. But it says it does not have the ability to learn, and it's up to the programmers to refine the way it works. Which is expected. I just wonder why it said it will strive to do better. It won't. It can't.
All lies and jest
Still a man hears what he wants to hear
And disregards the rest
-Paul Simon
User avatar
peter
The Gap Into Spam
Posts: 11488
Joined: Tue Aug 25, 2009 10:08 am
Location: Another time. Another place.
Been thanked: 4 times

Post by peter »

On a practical front, Fist, could you ask it why my shower drips when any tap in the house is turned on, upstairs or down? Is it (as I have been told) a problem with the 'shower bar' which likely needs replacing?

As good a test of its practical usefulness in everyday situations as you could find, I'm thinking.

;)
The truth is a Lion and does not need protection. Once free it will look after itself.

....and the glory of the world becomes less than it was....
'Have we not served you well'
'Of course - you know you have.'
'Then let it end.'

We are the Bloodguard
User avatar
wayfriend
.
Posts: 20957
Joined: Wed Apr 21, 2004 12:34 am
Has thanked: 2 times
Been thanked: 4 times

Post by wayfriend »

[...] In an economic race with enormous winner-takes-all stakes, a company is primarily thinking about whether to deploy their system before a competitor. Slowing down for safety checks risks that someone else will get there first. In geopolitical AI arms race scenarios, the fear is that China will get to AI before the US and have an incredibly powerful weapon — and that, in anticipation of that, the US may push its own unready systems into widespread deployment.

Even if alignment is a very solvable problem, trying to do complex technical work on incredibly powerful systems while everyone is in a rush to beat a competitor is a recipe for failure.
[link]
.
User avatar
Zarathustra
The Gap Into Spam
Posts: 19621
Joined: Tue Jan 04, 2005 12:23 am

Post by Zarathustra »

How Scientists Are Using AI to Talk to Animals

Now that’s cool!! This might be the way we communicate with aliens, too, should the need arise.

On rushing products to the marketplace before they are ready—isn’t this all software? That’s why we see a beta version and then endless updates. It hasn’t caused a catastrophe yet.

But whether the program works too well or doesn’t work well enough, some will always find reasons to fear. Doesn’t mean we should listen to them. It just means they have made up their minds to fear even when presented with opposite scenarios—in other words, the entire spectrum of possibilities. Is there an acceptable place between working too well and not working well enough where they aren’t going to predict doom and failure? Good luck getting them to define that Goldilocks zone.
Joe Biden … putting the Dem in dementia since (at least) 2020.
User avatar
peter
The Gap Into Spam
Posts: 11488
Joined: Tue Aug 25, 2009 10:08 am
Location: Another time. Another place.
Been thanked: 4 times

Post by peter »

I'm thinking that Alan Moore is probably right (see my post in another thread); that AI will probably remain a niche entity for a long time to come rather than be significant as a general 'across the board' effector within our personal lives (ie something that we use in place of our own faculties, and that hampers or causes retrogression of our development). In this sense it will already be in, or will move in very rapidly, to the goldilocks zone without our even being aware of it. How many of us give thought to the methods of say power distribution or traffic flow control that are operational behind the scenes as we go about our daily lives?
The truth is a Lion and does not need protection. Once free it will look after itself.

....and the glory of the world becomes less than it was....
'Have we not served you well'
'Of course - you know you have.'
'Then let it end.'

We are the Bloodguard
User avatar
Fist and Faith
Magister Vitae
Posts: 23439
Joined: Sun Dec 01, 2002 8:14 pm
Has thanked: 6 times
Been thanked: 30 times

Post by Fist and Faith »

I'm floored by the fact that bees listened to RoboBee!!! How can that, uh, be?? I thought they'd drive out bees from other hives. Based on pheromones? RoboBee night not have the "wrong" pheromones, since it probably doesn't have any. But, not having any, why do they recognize it as one of their own?
All lies and jest
Still a man hears what he wants to hear
And disregards the rest
-Paul Simon
User avatar
wayfriend
.
Posts: 20957
Joined: Wed Apr 21, 2004 12:34 am
Has thanked: 2 times
Been thanked: 4 times

Post by wayfriend »

peter wrote:I'm thinking that Alan Moore is probably right (see my post in another thread); that AI will probably remain a niche entity for a long time to come rather than be significant as a general 'across the board' effector within our personal lives
AI is in that class of things, along with robotics, telecommunication, computers, electronics, motor vehicles, etc. which change the labor market by reducing industry's dependence on human beings. Given how massively industry invested in telecommunication, computers, electronics, motor vehicles, etc. it is not unreasonable to see AI following the same path.

It won't be "niche". It's already in Alexa, Siri, etc.
.
Post Reply

Return to “General Discussion Forum”