Posted by SainSmart on

Written by Kitto

The world of CNC machining is rapidly evolving. CNC (Computer Numerical Control) machines are the workhorses of modern manufacturing, capable of creating complex parts with incredible precision. But who writes the instructions, the G-code, that tells these machines what to do? Traditionally, this has been the domain of human programmers. However, the rise of Artificial Intelligence (AI) raises the question: will AI eventually replace human programmers in writing G-code?

 

In the first video, the creator tests GPT-3's G-code programming capabilities by assigning it a simple task, such as facing a part or drilling four holes evenly spaced one inch apart. After ChatGPT gave him a program, he drew it using CIMCO Edit software to visualize the results and any issues in the code.

 

 

The Current State of AI and G-code

Large language models like ChatGPT are showing promise in assisting with G-code generation. These AI tools can understand natural language instructions and translate them into basic G-code commands. However, as the video we discussed demonstrates, there are limitations. AI-generated G-code often requires significant editing and correction by a human programmer to ensure functionality and safety. AI currently lacks the deep understanding of machining processes, physics, and machine capabilities that human programmers possess.

 

The Benefits of AI in G-code Programming

Despite the limitations, AI offers several potential benefits for G-code programming:

Increased Efficiency: AI can automate repetitive tasks, freeing up programmer time for more complex projects. Imagine an AI tool that automatically generates basic toolpathing for simple drilling or milling operations.
Reduced Errors: AI can be programmed to follow best practices and safety protocols, potentially reducing errors in G-code.
** Democratization of CNC Machining:** AI-powered tools could make G-code programming more accessible to those with less experience.

ChatGPT 4 can create G-code from images and PDF files. According to the article you provided, users can upload various file types to the interface, and ChatGPT will interpret the content. Users simply need to tell ChatGPT how to process the image. For example, if you upload a file of a part with four holes, you can tell ChatGPT to write a program to drill the holes. Wearne notes that this currently works best for simpler parts and shapes. "Anything beyond that is not feasible," he adds. "At least, not yet."

Wearne also notes that the G-code generated by ChatGPT has improved significantly. "It's as if GPT-4 can think more about its answers, whereas GPT-3.5 just gave out whatever answer it came up with as quickly as possible," he explains. Wearne says that with the latest update, it can program simple parts almost perfectly. GPT-3's code was graded as a high C or low B, while "for simple parts, if we use G-code 101, GPT-4 gets an A," he says.

 

 


The Role of the Human Programmer

While AI may become a valuable assistant, it's unlikely to replace human programmers entirely. Here's why:

Creativity and Problem-Solving: CNC machining is not just about following instructions. Programmers need creativity to solve problems and optimize machining processes for efficiency and quality.
Understanding the Bigger Picture: Programmers need to consider factors beyond G-code, such as material properties, machine limitations, and overall production goals.
The Human Touch: Experience and judgement are crucial. A programmer can identify potential issues in a design or G-code program that AI might miss.


The Future of Human-AI Collaboration

The future of G-code programming likely lies in collaboration between humans and AI. Imagine an AI tool that suggests different toolpathing options, helps identify potential collisions, and automatically generates basic code. The human programmer can then leverage this assistance to focus on the more complex aspects of the job.

 

In Conclusion

AI is a powerful tool with the potential to revolutionize G-code programming. However, it's not a replacement for human programmers. The future of CNC machining lies in collaboration, with AI augmenting human expertise to create a new level of efficiency and precision.

Back to blog

1 comment

Humans stopped writing GCode years ago. Everybody I know uses CAM because it is easier, better, and much much less error-prone. And that is just for CNC. For 3D printing, the GCode has never been written by humans; it is always done by a slicer.

William Bradley

Leave a comment

Please note, comments need to be approved before they are published.