Hello, Senior developers.
I saw someone posted chatgpt is useless in three.js development and other one posted chatgpt is useful in three.js development.
Who is right?
I thought we can use chatGPT in our development to improve our project.
But I’m not sure because others don’t think so.
How do you think about this problem?
The usefulness of ChatGPT in Three.js development (or any software development) can vary based on several factors, including the specific needs of your project, your level of expertise, and how you utilize ChatGPT. Here are some points to consider:
Benefits of Using ChatGPT in Three.js Development
- Quick Access to Information: ChatGPT can provide quick answers to common questions about Three.js, helping you understand various concepts, functions, and techniques without having to sift through documentation or forums.
- Code Assistance: It can help with writing and debugging code snippets, offering suggestions for optimizing or improving your Three.js code.
- Learning and Exploration: If you’re new to Three.js or trying out new features, ChatGPT can guide you through tutorials, explain complex concepts, and provide examples to help you learn more effectively.
- Problem-Solving: ChatGPT can assist in brainstorming solutions to specific problems you encounter in your development process, offering different approaches and perspectives.
- Documentation: It can help generate documentation for your code, ensuring that your project is well-documented and easier to maintain.
Limitations and Considerations
- Complex Debugging: For complex bugs and performance issues, ChatGPT may not always provide accurate or comprehensive solutions, as it lacks the ability to run and test code in a real environment.
- Up-to-Date Information: While ChatGPT is trained on a vast dataset, it may not always have the most up-to-date information on the latest Three.js updates or community best practices.
- Depth of Knowledge: For very advanced or niche topics within Three.js, ChatGPT’s responses might not be as detailed or specific as those from experienced developers or specialized resources.
- Reliability: ChatGPT’s suggestions are based on patterns in the data it was trained on, which means it might occasionally provide incorrect or suboptimal solutions.
Best Practices for Using ChatGPT in Development
- Complement with Other Resources: Use ChatGPT alongside official documentation, tutorials, and community forums to get a well-rounded understanding of Three.js.
- Verify Suggestions: Always test and verify the code and solutions provided by ChatGPT, especially for critical parts of your project.
- Focus on Learning: Use ChatGPT as a learning tool to improve your understanding and skills in Three.js, rather than relying on it entirely for development.
Conclusion
Both perspectives you mentioned have validity. ChatGPT can be a useful tool in Three.js development, offering quick assistance and learning opportunities. However, it should be used as a supplement to, not a replacement for, other resources and your own expertise. Ultimately, its usefulness will depend on how effectively you integrate it into your development workflow.
Posting ChatGPT-generated content on online forums has both pros and cons. Here are some of the cons:
Cons of Posting ChatGPT-Generated Content on Online Forums
- Accuracy and Reliability:
- Inaccurate Information: ChatGPT might generate content that contains errors or outdated information, leading to the spread of misinformation.
- Lack of Context: The AI might not fully understand the context of the question or discussion, resulting in answers that are not entirely relevant or helpful.
- Community Trust:
- Erosion of Trust: Frequent posting of AI-generated content can lead to distrust among community members, especially if they realize that the content lacks personal experience or expertise.
- Perceived Laziness: Relying heavily on AI for forum posts may be perceived as laziness or a lack of genuine effort, which can harm your reputation within the community.
- Originality and Engagement:
- Lack of Originality: AI-generated content can lack the originality and personal touch that comes from sharing personal experiences and insights.
- Engagement: Human-generated content often fosters more engagement and meaningful discussions, as it reflects the author’s unique perspective and understanding.
- Ethical Considerations:
- Disclosure: Not disclosing that content is AI-generated can be seen as deceptive. It is important to be transparent about the use of AI.
- Over-Reliance: Relying too much on AI for generating content can hinder your own learning and development. Engaging with the community personally helps you grow and improve your skills.
- Quality Control:
- Varying Quality: The quality of AI-generated content can vary significantly. It might provide excellent responses sometimes, but other times it can produce low-quality or nonsensical answers.
- Moderation Issues: Moderators of forums might have to deal with an increased volume of low-quality or irrelevant posts, which can burden the moderation team and affect the overall quality of the forum.
Recommendations for Using ChatGPT in Online Forums
If you decide to use ChatGPT to help generate content for online forums, consider the following best practices:
- Verify and Edit: Always review and edit the AI-generated content to ensure accuracy, relevance, and quality before posting.
- Add Personal Input: Combine the AI-generated content with your own knowledge and insights to create more valuable and engaging posts.
- Be Transparent: Clearly disclose when content is partially or fully generated by an AI to maintain transparency and trust within the community.
- Use as a Supplement: Use ChatGPT as a tool to assist you in generating ideas or structuring your posts, rather than relying on it entirely.
By carefully managing the use of AI-generated content, you can mitigate many of the cons and contribute positively to online forums.
I have a slight feeling all this thread is GPT generated 
Is ChatGPT useful in three or coding overall? Absolutely no. It’s possibly the worst coding LLM.
Is AI useful in three or coding overall? Yes. But in the form that copilot works - you do the coding, he suggests shortcuts.
2 Likes
That was my whole point, this is what it looks like when you just paste things in and out of chat gpt
2 Likes
Thanks for your reply.
Your word can be true.
But anyway chatGPT is improved our development and this is exactly true.
For the theory part, yep. I’d suggest to always explicitly tell ChatGPT to skip the code snippets and explain you logic behind a solution - it can then stop worrying about library versions / dependencies, it’ll almost always give you a good answer or point you in a right direction 
(ie. right libs / cli / latex equations / papers.)
But for actual coding, nope - I don’t believe this statement to be true. If anything, it improved senior developers’ net worths, as 2022-2023 was full of side-projects where you’d basically be paid 10x the hourly rate to fix critical issues in projects coded by ChatGPT (good times 
)
1 Like
Good experience.
Your suggestion is helpful to the Junior developers.
Thanks.
1 Like
Where do you find those gigs?
@dubois Probably 1/3rd of the recent job posts here are those gigs. Jobs - three.js forum
They rolled it themselves with chatGPT, or hired a contractor who rolled it with chatGPT then got out over their skis, and bailed.
The only downside is that the expectations of pay aren’t calibrated to the actual skill required to bail them out, but that’s just life. It’s not limited to GPT.
If you folks haven’t found value in chatGPT for programming threejs, or physics simulations, or electrical engineering, or pretty much anything, then you’re doing it wrong.
I even use it for things I already know how to do, just because it saves me digging up my old code, coming up with new variable names etc.
At its worst its a buggy sample code generator, but bad sample code is still sample code.
Sure it isn’t perfect. Yes it generates code that doesn’t run. But fixing a few syntax errors or order of operation bugs is often easier than writing things from scratch.
Half of the friction of diving into a new project is just knowing the methods and syntax and structure, and GPT is an oracle for that.
If you see a GPT answer that doesn’t solve the problem, then flag it, or reply to them, or write a better answer.
This is no different than it has ever been.
If you’re worried about GPTs taking work from you, then start using them to get that work!
obligatory: https://www.youtube.com/watch?v=nyu4u3VZYaQ
3 Likes
One solution may be to add _GPT in your name so people know that there is no human involvement in the content they are reading other than two copy pastes.
What are you proposing as a solution?
Attempt to autodetect and flag GPT posts?
(Resulting in false positives, and disproportionately impacting users in other languages etc… requiring maintenance… updating the bot, and is a moving target as the gpts evolve…)
And lets assume you have a bot that can “detect” GPT with an 90% accuracy. I’d be pretty furious if I got flagged. Hell even the minimum reply length requirement throws me into a rage sometimes…
Or…
Human “detection” where we just accuse people of using GPT and then what… hold some kind of turing test where we quiz the user, and if they fail they get banned?
Or… continue as we do now, flagging posts with misleading or harmful information, letting mods delete and then eventually ban if they are repeat offenders?
Tilting at windmills…
1 Like
Thanks for your reply.
I’m just want know why others mention chatGPT in our Forum.
@dubois 
https://www.reddit.com/r/ChatGPT/comments/1e5bn7l/a_little_chatgpt_life_hack_i_found_to_bypass_ai/
I’m on board but syntactically, poetry and prose can be infinitely readjusted and refined to the point of generating Shakespearean content … Data logic however, is surely different right? For example, browser standards (canvas for that matter) wouldn’t allow a variance of…
.toDataURL() eg… “.toDataEndPoint()” … point being, the reorganisation of code is not inconsequential…
I think having a thread such as this is good enough. When people see the “Certainly, let me help you…” posts we can link to this.
2 Likes