ChatGPT can write in assembly language

[LiaoMi]

  Hi,

AI successfully generates some examples!  Has anyone tested other features of this neural network? 

[Biterider]

Hi LiaoMi
http://masm32.com/board/index.php?topic=10004.msg116831#msg116831

Its capabilities are really impressive. I looked at it again last week, I noticed that it has learned in the meantime.

Yesterday, I talked to an IT professor at the University of Konstanz about what the new strategy for dealing with AI will be. They don't want to ban the use of ChatGPT, which is practically unfeasible, but they do want to push the use of AI for coding and for students to learn how to properly ask questions of AI.

Times are changing fast and this technology is certainly disruptive.

Biterider

[C3]

I asked ChatGPT for 64bit assembly code with specific parts of code, but it didn't succeed, just gave a randon error code.
Well the program wouldn't work ok.

[caballero]

Hi, just to contrast

Futurists insist that AI will soon eclipse the capacities of the most gifted human mind. What hope do we have against superintelligent machines? But we aren’t really on the path to developing intelligent machines. In fact, we don’t even know where that path might be.

A tech entrepreneur and pioneering research scientist working at the forefront of natural language processing, Erik Larson takes us on a tour of the landscape of AI to show how far we are from superintelligence, and what it would take to get there. Ever since Alan Turing, AI enthusiasts have equated artificial intelligence with human intelligence. This is a profound mistake. AI works on inductive reasoning, crunching data sets to predict outcomes. But humans don’t correlate data sets: we make conjectures informed by context and experience. Human intelligence is a web of best guesses, given what we know about the world. We haven’t a clue how to program this kind of intuitive reasoning, known as abduction. Yet it is the heart of common sense. That’s why Alexa can’t understand what you are asking, and why AI can only take us so far.

Larson argues that AI hype is both bad science and bad for science. A culture of invention thrives on exploring unknowns, not overselling existing methods. Inductive AI will continue to improve at narrow tasks, but if we want to make real progress, we will need to start by more fully appreciating the only true intelligence we know―our own.

https://www.amazon.es/Myth-Artificial-Intelligence-Computers-Think/dp/0674983513/ref=asc_df_0674983513/?tag=googshopes-21&linkCode=df0&hvadid=469663859267&hvpos=&hvnetw=g&hvrand=9042876580470242797&hvpone=&hvptwo=&hvqmt=&hvdev=c&hvdvcmdl=&hvlocint=&hvlocphy=1005491&hvtargid=pla-981582275036&psc=1

[hutch--]

I agree with that conclusion, use what you started with and keep improving it. The only thing AI has is the first initial, "artificial" but no intelligence.

Interestingly enough, it is easy enough to emulate what the current AI happens to be but on a far smaller scale. Start with a simple tree, keep adding nodes to it in both width and depth with each node storing data and to increase its range, make it possible for each node to link to any other node to get data from that location. With new ideas (pseudo learning) make it possible to create new trees that have the same capacities as the original tree of being able to link to any other node and you are on you way to creating a pseudo intelligent piece of software.

The difference between a tiny project like this and what is being peddled as AI is only the scale available to search engines, a truly massive volume of links to many subjects.

[jj2007]

The only thing AI has is the first initial, "artificial" but no intelligence.

Intelligence is the ability to combine bits and pieces of information so that new information is created that helps to solve problems (my personal, spontaneous interpretation of the phenomenon). How exactly that happens, in a brain or a cpu, is worth studying but not really that important - the result is more interesting. We are at a very early stage of AI, but ChatGPT or Character.AI can already compete with a great share of the population.

Btw we are debating here ("ChatGPT can write in assembly language") a skill that is mastered by how many % of the population? 0.001%? Would you argue that 99.999% of the population don't show signs of intelligence, because they can't write Assembly code?

jj: How intelligent was your decision to buy twitter?
Elon: I think it is one of the most intelligent things I've ever done. In my opinion, Twitter is extremely important to modern society. There's so much to say on the topic, but I don't want to drone on for too long. The main point is that Twitter is one of the most powerful assets in human history, and it was being criminally underutilized.

jj: Was bringing Donald back a good idea, in hindsight?
Elon: I think bringing Donald back was a bad idea. I believe he is very divisive, and I think that division is extremely unhealthy for the human mind. I don't think Twitter was made for people to yell at each other. I want Twitter to be a place where people can have civil conversations, and try to understand each other, not insult and demean each other.

[hutch--]

> but ChatGPT or Character.AI can already compete with a great share of the population.

A computer can always regurgitate data faster than a human but get it to do something intelligent.

[jj2007]

So you can't see any difference between "regurgitate data" and the Musk bots answers to my questions above?

[greenozon]

Any examples (screens?) of what ChatGPT is capable of for x64 MASM please ;)

[JonasS]

I did not know this!? 

Me: who owns masm32?
ChatGPT: The MASM32 toolkit is owned and maintained by a software developer named Hutch-- (real name: Randall Hyde), who is also the author of "The Art of Assembly Language Programming", a widely used textbook on assembly language programming. Hutch-- first released MASM32 in 1997 and has continued to update and improve it over the years. The MASM32 toolkit is distributed as freeware and is widely used by assembly language programmers, particularly those developing software for the Windows platform.

 

[hutch--]

 

[HSE]

   A big future in politics.

[jj2007]

Randy, they caught you in the act

[Biterider]

Hi
For those interested in using GPT, here is a good starting point
https://www.codeproject.com/Articles/5355950/ChatGPT-API-for-Windows-Use-ChatGPT-in-your-Window

I failed using the code as it is, but I gave it a try using

Code: [Select]

c.Image(input.c_str())and I successfully got an image associated with the input prompt.

With a little work it should be possible to use any GPT model from our asm programs 

Biterider

[ewok]

I asked it to code a software that write hello world into a file and verify before if the file don't contain already 'hello world' in masm32
It did recognize what I asked but have outputed bad codes (forgot some include since for some reason it used C apis, .data sections things like that, then tried to code review it, it kinda came back with some corrections after, but it's like I have to rewrite most of the thing myself.
It creates things that look like what the code would be but doesn't work.

Meanwhile, if I ask everyday things, many times it's like wikipedia or something, not very interesting. Then I got bored..
I don't know how it evolved now, but I don't think it will replace a masm programmer.