Introduction

Hey there, fellow open source developers! Today, we need to have a serious talk about AIs like Claude, ChatGPT and Lamma. It’s time to start learning and using them – for real. These AI models have learned from our code, and it’s high time we benefit from this technology too!

Now, I know what some of you are thinking, and let’s address the two big concerns head-on:

“But it’s not open source!”
“Will AI replace me or make my skills useless?”

Let’s tackle these one by one.
Open Source Concerns
You’re partly right about the open source issue. Many of the most powerful AI models are indeed proprietary. However, the open source community is catching up fast. Take LLaMA 3.2, for instance. It’s a really powerful model – not quite as good as Claude, but way better than nothing. And it’s open source. It’s very hard to run the 405B model localy but you can use a paid service.
The field is evolving rapidly, and more open-source alternatives are emerging every day. As open source developers, we have the opportunity to contribute to and shape these tools, ensuring they align with our values and needs.
Now, let’s address the bigger elephant: the fear of being replaced or feeling like your skills are becoming obsolete. Let me be clear: AI is not here to replace you – it’s here to amplify your capabilities.
Think of AI as a powerful tool in your developer toolkit. Just like how we learned to use IDEs, version control, and cloud platforms, AI is the next evolution in our field. Your skills, creativity, and problem-solving abilities are more valuable than ever. AI can handle the repetitive tasks, but it needs your expertise to guide it, to understand the bigger picture, and to create truly innovative solutions.
By embracing AI, you’re not rendering your skills obsolete – you’re evolving them. You’re becoming a developer who can leverage AI to work more efficiently and tackle more complex problems. The developers who adapt and learn to work alongside AI will be the most valuable in the coming years.
So yes, the field is changing rapidly. Open-source AI alternatives are emerging, and the landscape is evolving. But this isn’t a threat – it’s an opportunity. An opportunity for us, as open source developers, to be at the forefront of this revolution, shaping how AI is used in software development.
Remember, open source has always been about collaboration, innovation, and democratizing technology. By engaging with AI, we’re continuing that tradition, ensuring that these powerful tools remain accessible and aligned with open source principles.

With that out of the way, let’s cover some misconceptions.
“programming in English.” It doesn’t mean quite what people think it means – at least not yet. Most of what I do now when I code is type in English, but my requests look something like this:

“Use PyTorch, structure like the attached script, use the GitHub page about Whisper to do ASR. The script will be a local AI system that analyzes text files in a folder and saves them to ./sum.txt”

Effective AI Prompting

The key is to write your message as if it was meant for a developer you want to follow specific tasks. This is how you get well-structured code quickly. You’ll often need to copy in GitHub READMEs and relevant documentation. This means you need to be able to locate and read documentation effectively.

Handling Errors

Now, let’s talk about errors. When you’re working with AI, you need to read the errors and understand them. If the AI is reiterating errors, that’s actually a good thing – it’s trying to help you debug. But if things keep repeating the same error, you need to fix it yourself. It does not mean the AI is bad, it means you need to help it get unstuck from it’s local minimum. Once you fix the error yourself, you can always copy the updated code back in and keep working.

Sometimes, you can start a new chat, copy in the relevant files, and describe the issue to get unstuck. But in many cases, it’s important to be able to fix the issue on your own. This is where your coding skills really come into play.

The Real Impact of AI on Coding

So, can AI be used for coding? Hell yes! But here’s the kicker – it’s most useful to those who can already code fairly well. For me, it took me from being able to code anything I want, to being able to code anything I want stupidly fast.

Conclusion

In conclusion, AI is not replacing us as developers. It’s augmenting our abilities, making us faster and more efficient. But to really harness its power, we need to understand coding fundamentals, know how to read documentation, and be able to troubleshoot effectively.

So, my fellow open source devs, don’t fear AI – embrace it. Learn it. Use it. And let’s push the boundaries of what we can create together. The future of open source development is here, and it’s powered by AI. Are you ready to be part of it?


Leave a Reply

Your email address will not be published.