chatgpt

A.I. could be usefull when i no longer know the right pronoun.
The drunk uncles had read a poem by chatgpt about that topic this week. It was good, but politically very incorrect. That's why I cant link here.
 
You'll notice very quickly the beginning of every answer is constructed from your question.

See misc/p5-Chatbot-Eliza

"ELIZA's creator, Weizenbaum, regarded the program as a method to show the superficiality of communication between man and machine, but was surprised by the number of individuals who attributed human-like feelings to the computer program, including Weizenbaum's secretary. Many academics believed that the program would be able to positively influence the lives of many people, particularly those with psychological issues, and that it could aid doctors working on such patients' treatment."

Nothing new in 60 years (?)(!)

https://en.wikipedia.org/wiki/ELIZA
 
What if ChatGPT could get angry enough to act irrationally? Can it own up to that? About a year or so ago, there was this colorful guy (Trihex) on these forums, he had a ball with chatbots. I bet that guy could get ChatGPT pissed at him if he tried hard enough. 🤣
 
Has anyone played around with chatgpt, thoughts?

I used it a bit and it is pretty impressive in terms of being able to provide relevant answers to a broad range of topics and perform various tasks. As a software engineer by day, I am both scared and excited of the future. Even my role could be outsourced to AI it appears.

I gave it a go and was half impressed. For my first question ("how can I disable the pthread stack cache on FreeBSD") it gave a very clear, well explained and plausible answer. The only problem was that the answer was wrong (it suggested modifying the pthread attribute for the stack size, but FreeBSD pthreads have two stack caches, one for default sized thread stacks and one for user sized). I was looking more for a hack than just abusing the public API, and chatgpt just discouraged using hacks. My second question was about macOS mach messages and why the kill message might not kill a thread immediately. This time chatgpt didn't have much of a clue and waffled a bit about scheduling.

So my conclusing is that the back end for building up clear explanations is very impressive. The front end (the actual "AI") is fairly underwhelming.
 
My second question was about macOS mach messages and why the kill message might not kill a thread immediately. This time chatgpt didn't have much of a clue and waffled a bit about scheduling.
When it comes to process scheduling, that's actually a pretty important concept in OS design. Different OS'es have somewhat different implementations, but largely the same goals. Those goals include OS stability and perceived speed. Depending on the implementation, you can't just remove a running thread from RAM willy-nilly. I'd say ChatGPT had a good start there.
 
When it comes to process scheduling, that's actually a pretty important concept in OS design. Different OS'es have somewhat different implementations, but largely the same goals. Those goals include OS stability and perceived speed. Depending on the implementation, you can't just remove a running thread from RAM willy-nilly. I'd say ChatGPT had a good start there.
If it assumed that I was using OS scheduling then that is a bad start. I tried again explaining that I'm using cooperative multithreading with a single global lock and the suggestions were better. I just want to be able to do the same thing that libpthread does internally when a thread finishes.
 
For 40 years I have been hearing how CNC Machines are going to replace machinists.
All hype.
Not everything in manufacturing is new construction or mass produced.

OpenAI chatgpt is not revolutionary. It is devolutionary.
Agree: I feel it contributes to the "stupefying" (stupidifying?) of the human race. People generally don't want to think and just want to ask a question and get an answer.
 
If it assumed that I was using OS scheduling then that is a bad start. I tried again explaining that I'm using cooperative multithreading with a single global lock and the suggestions were better. I just want to be able to do the same thing that libpthread does internally when a thread finishes.
So, basically, you want to use your own code to override the OS's native implementation of thread management (libpthread)and be able to 'kill' a thread faster that way?

Nothing wrong with that, but it would take a lot of setup to make sure that research doesn't bring down the very machine it's running on. I think that pulling it off successfully would take respecting the natively implemented limits and process scheduling of the OS.
 
Code:
some code
Code:
some code
So, basically, you want to use your own code to override the OS's native implementation of thread management (libpthread)

Yes.

and be able to 'kill' a thread faster that way?

No. It's not about speed (I wish!). It's about having total control.

I need to look at the XNU source to see more what is going on.

pthread_create does it's stuff (allocates memory that is used for the stack, creates some sort of LWP to run the thread. thread_start calls the user function and then on return it calls thread_exit.


Code:
* thread #2, stop reason = instruction step into
    frame #0: 0x00007ff81665d259 libsystem_pthread.dylib`_pthread_start + 125
libsystem_pthread.dylib`:
->  0x7ff81665d259 <+125>: movq   %rbx, %rdi
    0x7ff81665d25c <+128>: movq   %rax, %rsi
    0x7ff81665d25f <+131>: callq  0x7ff81665d8b7            ; _pthread_exit
    0x7ff81665d264 <+136>: movl   %r9d, %eax

There's a lost of stuff in pthread_terminate and pthread_tsd_cleanup but eventually we get to

Code:
* thread #2, stop reason = instruction step into
    frame #0: 0x00007ff81665b527 libsystem_pthread.dylib`_pthread_terminate + 490
libsystem_pthread.dylib`:
->  0x7ff81665b527 <+490>: callq  0x7ff816660728            ; symbol stub for: __bsdthread_terminate
    0x7ff81665b52c <+495>: leaq   0x5886(%rip), %rcx        ; "BUG IN LIBPTHREAD: thread didn't terminate"
    0x7ff81665b533 <+502>: movq   %rbx, %rax
    0x7ff81665b536 <+505>: movq   %rcx, 0x41d60f53(%rip)    ; gCRAnnotations + 8

Where bsdthread_terminate is a syscall, 0x2000169.

That seems to do it's job.

I'm still none the wiser as to why the mach kill message didn't work. I'll try an old OS X to see if it is different.

And back to ChatGPT - it really is not capable of this sort of detailed analysis.
 
Back on any thoughts about ChatGPT and other artificial text generators, this article might be interesting food for thought:

"AI systems should not pretend to be providing neutral and factual information while displaying clear political bias"

`Artificial Text Generators®`, since there is no intelligence involved, unless it comes from the CIA

Tried to try this 2023 version of Eliza, but the chat.openai.com website doesn't seem to understand my www/luakit browser. "Please stand by, while we are checking your browser...". The command to 'verify' if I'm human is also not very instructive (pinched myself, but the site didn't respond).
chat-openai-com_human-verification.png

At least this chatbot is a good laugh.
 
And back to ChatGPT - it really is not capable of this sort of detailed analysis.
I frankly didn't expect it to be... but this kind of analysis is something that ChatGPT can be taught. Average IQ of AI implementations has risen over time... when I was in college (2000-2006), AI was in its infancy. Even with robotic adherence to logic formulas, average IQ of AI implementations rivaled that of human toddlers. These days (more than 15 years down the road), it can outperform an average human being on SAT. No, you can't rely on it for expert analysis yet, it needs adult supervision, just like humans. Case in point: Sam Bankman-Fried. I thought he's a smart kid who just didn't quite connect the dots, and ended up acting recklessly, without thinking too far ahead.
 
Has anyone played around with chatgpt, thoughts?

I used it a bit and it is pretty impressive in terms of being able to provide relevant answers to a broad range of topics and perform various tasks. As a software engineer by day, I am both scared and excited of the future. Even my role could be outsourced to AI it appears.
A lot. And here is yet another - Writesonic.

What is really amazing it that it actually learns over time.

This Writesonic seems to be connected to the real internet and updates its information. Here is what I asked:

Code:
Please write a detailed summary of https://forums.freebsd.org/


FreeBSD Forums is a community-driven discussion forum for users of the open-source FreeBSD operating system. It provides an opportunity for members to post questions, share experiences and resources, and discuss topics related to FreeBSD. The forums, which have been around since 2006, are divided into several categories, including General Discussion, Documentation and News, Development, and Hardware, Networking and Security. There is also a special area dedicated to FreeBSD-based operating systems, such as PC-BSD and GhostBSD. The forums offer a wealth of information, from installation instructions to troubleshooting tips. Additionally, members can use the forums to request help from other users and from FreeBSD developers. The forums also provide a platform for people to share their experiences, discuss ideas, and collaborate on projects.

Another chat:

Code:
Describe the most important characteristics of FreeBSD in detail!

FreeBSD is a free and open-source Unix-like operating system based on the Berkeley Software Distribution (BSD). It is one of the most widely used operating systems for web servers, due to its robustness and stability.

The core of FreeBSD is a complete operating system kernel, which manages the hardware and software resources of the system, and provides essential services such as process scheduling, memory management, and inter-process communication. It provides a set of APIs to facilitate the development of a wide range of software, from system utilities to graphical user-interfaces.

Additionally, FreeBSD contains a package management system that includes a wide range of packages and tools, such as compilers, interpreters, database servers, web servers, and even windowing systems. The package management system also allows software to be updated easily and quickly.

Finally, FreeBSD provides an extensive collection of software that allows developers to create applications quickly and efficiently. This includes a wide range of open source libraries and components, as well as a variety of development tools such as compilers, interpreters, and debuggers.

In conclusion, FreeBSD is an extremely powerful and reliable operating system that offers a wide range of features and tools for developers. It is highly extendable, secure, and stable, making it a popular choice for both desktop and server applications.

And the last one:

Code:
What is the future of FreeBSD?


The future of FreeBSD looks bright, with the FreeBSD Foundation actively investing in development projects to improve the OS and its capabilities. [1] This includes WiFi improvements, OpenJDK, ZFS RAID-Z expansions, security, toolchain, and performance improvements. [2] Many companies, including Netflix, rely on FreeBSD for its reliability and stability, making it well-suited for desktop use. [3] Despite the challenges posed by slower reviews of updates and support for new hardware, Project Trident is one of many distros available to the desktop user.
 
A lot. And here is yet another - Writesonic.

What is really amazing it that it actually learns over time.

This Writesonic seems to be connected to the real internet and updates its information. Here is what I asked:

Code:
Please write a detailed summary of https://forums.freebsd.org/


FreeBSD Forums is a community-driven discussion forum for users of the open-source FreeBSD operating system. It provides an opportunity for members to post questions, share experiences and resources, and discuss topics related to FreeBSD. The forums, which have been around since 2006, are divided into several categories, including General Discussion, Documentation and News, Development, and Hardware, Networking and Security. There is also a special area dedicated to FreeBSD-based operating systems, such as PC-BSD and GhostBSD. The forums offer a wealth of information, from installation instructions to troubleshooting tips. Additionally, members can use the forums to request help from other users and from FreeBSD developers. The forums also provide a platform for people to share their experiences, discuss ideas, and collaborate on projects.

Another chat:

Code:
Describe the most important characteristics of FreeBSD in detail!

FreeBSD is a free and open-source Unix-like operating system based on the Berkeley Software Distribution (BSD). It is one of the most widely used operating systems for web servers, due to its robustness and stability.

The core of FreeBSD is a complete operating system kernel, which manages the hardware and software resources of the system, and provides essential services such as process scheduling, memory management, and inter-process communication. It provides a set of APIs to facilitate the development of a wide range of software, from system utilities to graphical user-interfaces.

Additionally, FreeBSD contains a package management system that includes a wide range of packages and tools, such as compilers, interpreters, database servers, web servers, and even windowing systems. The package management system also allows software to be updated easily and quickly.

Finally, FreeBSD provides an extensive collection of software that allows developers to create applications quickly and efficiently. This includes a wide range of open source libraries and components, as well as a variety of development tools such as compilers, interpreters, and debuggers.

In conclusion, FreeBSD is an extremely powerful and reliable operating system that offers a wide range of features and tools for developers. It is highly extendable, secure, and stable, making it a popular choice for both desktop and server applications.

And the last one:

Code:
What is the future of FreeBSD?


The future of FreeBSD looks bright, with the FreeBSD Foundation actively investing in development projects to improve the OS and its capabilities. [1] This includes WiFi improvements, OpenJDK, ZFS RAID-Z expansions, security, toolchain, and performance improvements. [2] Many companies, including Netflix, rely on FreeBSD for its reliability and stability, making it well-suited for desktop use. [3] Despite the challenges posed by slower reviews of updates and support for new hardware, Project Trident is one of many distros available to the desktop user.
Not bad, how about asking ChatGPT what are the downsides of using FreeBSD as opposed to other OS'es? Seeing all this uptalk makes me think ChatGPT is being fed propaganda. I'm baiting for stuff like wifi card, bluetooth and GPU support. Will ChatGPT be able to fess up to that?
 
On a recent episode of Security Now, Steve Gibson mentioned that a listener had used ChatGPT to create a PowerShell program to decode last pass data - including a GUI. Particularly that last part is quite astonishing. It wasn't perfect but the fact, that it managed to pull this out of "thin air" (so to speak, yes, there's a a bunch of "non-air" pushed in the back) is mind boggling - considering the underlying mathematics.

We're throwing a bunch of computing power at very open ended problems and the major issue with artificial intelligence remains: we're getting results that seem right but become almost impossible to verify and we don't fully understand how the thing got to it's conclusion.

On top of that, consider all the bias that was introduced using internet sources... as if today's internet was free of disinformation and partisan political self-serving interests. We're getting the self-serving echo chamber multiplied by computing power that will never be achieved with human bot-farming. We won't be able to tell truth from fiction in the worst case scenario. At least for now, it seems the powers to be have managed to moderate away the feedback loop problem we saw from "Tay" and similar endeavors. Zero to Hitler in 16 hours...

Time to read up on Paul Watzlawik again - how real is real?
 
according to Douglas Adams, it's the lawyers that'll be the first to go, if I remember correctly... not sure if it was in reference to a machine army rising, though.
This is funny, and I believe it. Some time ago I asked ChatGPT to compare two similar laws in two different countries and the comparison was a good one...

Also, what comes to Douglas Adams, I asked ChatGPT to generate a random number. It did, but later, when I asked how did it come to that answer, it denied the ability to generate random numbers. Then I asked to generate a certain fixed number if it cannot generate random numbers, and it replied with 42.

I asked (not in English and therefore I am not making a direct copy here) what is 42 and it denied first any connection and replied that it is 6 * 7. But some time later it confessed that:

Code:
According to Douglas Adams' book "The Hitchhiker's Guide to the Galaxy." It is a fictional number used to represent the ultimate answer to a question of ultimate importance, and is not a real-world mathematical concept.

So it made a joke.

Some time later I asked for random numbers again and it generated these without any problem in a given range and even told that it has a built-in PRNG...

EDIT: And I just asked about lawyers. It gave a very diplomatic answer to my question:

Code:
As an AI language model, I don't have personal opinions or beliefs.
But it is likely that AI will have a significant impact on many professions, including the legal profession. The use of AI in the legal field can help increase efficiency, accuracy, and accessibility of legal services.
However, it is important to remember that AI is a tool, and like any tool, it can only be as effective as the humans who use it.
Therefore, it is important for lawyers to continuously improve their skills and adapt to the changing landscape of the legal industry in order to remain relevant in an increasingly AI-driven world.
 
Just saw a question on Stack Overflow where they said they were looking for a solution to a problem. So they asked ChatGPT but the solution offered didn't work and they wanted to know why.
Too soon, I think. I feel doing such a thing is stooping too low, too, and I feel we'll start getting a rash of questions from people complaining about why the "chatgpt solution" didn't work.
 
But that is the problem right there: if you ask somebody else (for example chatGPT) to do some work for you, and you lack the knowledge (or resources) to verify that the result is correct, then you are back to square one again.
Soon somebody will suggest that you use Google AI to verify chatGPT answers (or vice versa) then everything related to this new AI craze will just be shit. as in "shit in, shit out".
The current level of AI's are good mimics, but they utterly lack any understanding of context.
 
Back
Top