# ChatGPT for Programming Numerical Methods

**URL:** <https://fortran-lang.discourse.group/t/chatgpt-for-programming-numerical-methods/6047>\
**Category:** AI\
**Created:** [June 27, 2023, 3:03am UTC](https://fortran-lang.discourse.group/t/chatgpt-for-programming-numerical-methods/6047 "2023-06-27T03:03:30Z")\
**Posts on this page:** 1\
**Showing post:** 23

<div class="post-metadata">

**Author:** ![R\_cubed](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/r_cubed/32/263_2.png) [@R\_cubed](https://fortran-lang.discourse.group/u/R_cubed)\
**Post date:** [August 20, 2024, 4:13pm UTC](https://fortran-lang.discourse.group/t/chatgpt-for-programming-numerical-methods/6047/23 "2024-08-20T16:13:51Z")

</div>

I’d expect any code generated by LLMs to be very fragile at best, and seriously wrong at worst, especially in the context of numerical methods. The reasons were discussed in this thread:

> [@Are we on the verge of new disruptive AI technology?](https://fortran-lang.discourse.group/t/are-we-on-the-verge-of-new-disruptive-ai-technology/7652/7):
>
> I don’t understand why any of this is terribly impressive. At the dawn of computing, it was easy to simulate human speech using pattern matching and some built-in knowledge of context. People would develop strong reactions to the exchange, and find it hard to believe it was not another person. He (inventor Joseph Weizenbaum) was surprised and shocked that individuals, including Weizenbaum’s secretary, attributed human-like feelings to the computer program The only “advance” today is th…

Specifically, this post by AI scholar Gary Marcus illustrates the fundamental limitations of stochastic prediction as an approach to knowledge.

> **[“Math is hard” — if you are an LLM – and why that matters](https://garymarcus.substack.com/p/math-is-hard-if-you-are-an-llm-and)**
>
> No matter how much data you train them on, they still don’t truly understand multiplication.

> That, in a nutshell, is why we should never trust pure LLMs; even under carefully controlled circumstances with massive amounts of directly relevant data, they still never really get even the most basic linear functions.

---

_[View the full topic](https://fortran-lang.discourse.group/t/chatgpt-for-programming-numerical-methods/6047)._
