Re: ChatGPT refutes the key rebuttal of my work

On 10/14/24 11:38 AM, olcott wrote:

On 10/14/2024 6:21 AM, Richard Damon wrote:

On 10/14/24 5:49 AM, olcott wrote:

On 10/14/2024 4:04 AM, Mikko wrote:

On 2024-10-13 12:53:12 +0000, olcott said:
>

Although it is possible for LLM systems to lie:
https://en.wikipedia.org/wiki/Hallucination_(artificial_intelligence)
>
ChatGPT does correctly apply truth preserving operations to
the premises that it was provided regarding the behavior of
DDD and HHH. *Try to find a mistake in its reasoning*

>
No reasoning shown.
>

>
https://chatgpt.com/share/6709e046-4794-8011-98b7-27066fb49f3e
>
When you click on the link and try to explain how HHH must
be wrong when it reports that DDD does not terminate because
DDD does terminate it will explain your mistake to you.
>
>

>
No, it admits that DDD does halt, but that HHH must be correct to say it doesn't, ... because of the lies you told it.
>

 It proves that it has a much deeper understanding than
anything that I told it.

Nope, just that it has seen a lot of arguments like yours and thus knows what you are looking for.
Remember, Chat GPT has been trained to be a "Yes Man"

 

Its reasoning is based on the incorrect presumption that the HHH that DDD calls is not part of the program DDD,

 (1) DDD never has been a program it is a C function.

Then you can't talk about its "Halting Behaivor" as that is a property of a PROGRAM.

(2) HHH does correctly emulated itself emulating DDD
     this <is> a contiguous sequence of computation.

Except YOUR HHH changes its behavior based on a non-input, and thus you claim is demonstrated to be a LIE.
Also, it DOESN'T "correctly" emulate itself per the definition that determines halting, as it stops before that answer was actualy determined, as shown by the fact that DDD() returns when HHH(DDD) says it doesn't.
Your logic is just based on lies.

 

because you have broken the definition of a program.

 I am not the one saying that a C function <is> a program.
You should not be so sloppy in your use of terminology.

But functions CAN be programs, when you include the code (if any) that they call. Since you claimed to be determining the halting status of DDD< that becomes a necessary condition to do so, that you ignore.

 DDD emulated by HHH including HHH emulating itself
emulating DDD is a contiguous sequence of computation.

Except that your HHH changes its behavior, and thus makes itself NOT a computation.
It also doesn't complete the task, so it never learns the behavior that it is supposed to answer. Just because you code computes something, doesn't make the answer correct for the problem actually being asked.
Not understanding that question really helps in making that sort of error.

 It is not and never has been a program. I think of
DDD and HHH as virtual machines.
 

Then you have been lying all this time, as termination is strictly a property of PROGRAMS, not functions that are not competely specified because they use code outside of what has been specified.
Sorry, you just sunk your ship.