LLM Showdown: ChatGPT and Claude Tackle STM32 Push Button LED Toggle
Вставка
- Опубліковано 2 лип 2024
- In this exciting installment of our AI Challenge series, we pit ChatGPT against Claude Sonnet to see which large language model can better handle programming an STM32 ARM microcontroller. The task? Using a push button to toggle an LED. Watch as we guide both AI models through the process, comparing their approaches, efficiency, and overall performance. Who will come out on top in this battle of artificial intelligence? Tune in to find out and gain insights into using LLMs for microcontroller programming!
Get the Newbiehack Kit:
newbiehack.com/Categories/ARM
My ARM Programming Book (Vol 1): amzn.to/4cPjGzB
STM32F030 GPT: chatgpt.com/g/g-848nlS6JC-stm...
Initial setup:
• #2 ARM Microcontroller...
• #3 ARM Microcontroller...
How to Make Multiple Pin External Interrupts (2 pins) - ARM STM32 Microcontroller Tutorial
• 55. How to Make Multip... - Наука та технологія
Thank you all for watching. I hope you enjoy and stay tuned for more videos. I just put together a camera jib, so I won't be tripping over my tripod anymore, which was a major reason for not doing videos for a little while. I just couldn't get the idea of bumbling over awkward situations with my camera tripod anymore. I might post a video on what the camera jib looks like and what went into the design, so let me know if you are interested in seeing this.
very good
From iraq🇮🇶.... Thank you for this informative video
Glad you like it! Thanks!
If one can not code a very simple io-function without chatgpt then maybe embedded coding isn't what one should do.
This is a series that test LLMs and each video is a little more difficult than the last (scaffolding). Just like the content, the series is also fledgling.
There are pro's and cons of AI's. It makes people very lazy at coding. Because the AI can write lots of code and knows so much.
For the fun I took my old unfinished projects and gave some instructions to ChatGPT 4o and Claude to make me some functionality. Both for C++ and Delphi (Pascal). And they generated me working code and even examples and test. To verify if the code is bug free. They even assisted with bugs and fixed them. Now i have working code for specific parts but I have no idea how it works and why it works. AI helps on this too but this makes me lazy.
It even generated me code that I was looking for almost 10 years. I couldnt make it. And AI made me this code and even made full example app in just 5 minutes total (with bug fixes and reviews) and it works.
Very true. The coding issues of us needing to understand the code will be short-lived, in my opinion. The acceleration of LLMs with getting better at coding is going so fast that in a year, or a few years, it will make coding for mere mortals obsolete. I believe that creativity will still be needed by humans for at least a few more years. We are in a strange territory in history.
Be wary of trusting AI to generate unit tests. If it’s generating any incorrect logic in the code to be tested, chances are it wrote the test with the incorrect expectations and can lead you to a false sense of security.
@@BigfootPrinting Absolutely. At least, curently.
@@BigfootPrinting Funny, it looks like humans are as good as AI at finding bugs and this, I'm sure, will relate to unit tests in the future: ua-cam.com/video/cw9jcjpufBI/v-deo.html
@@PatrickHoodDaniel Interesting, thanks for sharing that!