Just an emerging tech that folks are trying to wrap their heads around. No need to kill the thread, LLMs have thier purposes and understanding what that is will continue to evolve just as the tech does.
4 Likes
You can’t. Only bossman and jeffe.
2 Likes
It’s really hard to read my tone from my text. But I am not offended and not bothered by the subject at all.
I feel pretty strongly that I don’t want people to answer questions here by just prompting AI. Mostly because it feels too similar to “let me google that for you” or “read the F-in manual”. If someone wants to ask AI, they can already without the forum.
However, I am not bothered by you posting about it or the discussion so far. I’m reading your tone and you seem to be walking carefully, trying not to offend me. I just want to make it clear I am not offended and I’m sorry if I made it seem like I was.
I can close it if you want. But I like seeing the discussion and I like to use the close topic carefully.
5 Likes
I also like seeing the discussion, and I vote to keep the thread open.
2 Likes
Thanks, Jeff. We’re good. I think my original title was a bit too general about asking AI questions… and making it more specific to what I actually asked it was probably a step in the right direction IMO.
I think we’re all in agreement that we should always be skeptical of what AI tells us… and that there are times/topics when using it might be helpful. It appears most think I was using it in my first post in a way that could be useful. Certainly I thought it was/is and just wanted to share my excitement… thinking others who might have tried these programs and tried to do v-carve inlays could benefit from the information also. For me it was the terminology and settings differences that caused me some uncertainty and confusion… and I thought Gemini did a pretty good job of helping to clear my confusion.
I guess I’d like to see more discussion of where AI could be used in a way that might be helpful… rather than a continual string of responses about the dangers of AI and how we must always be skeptical and untrusting of what it tells us. I think asking AI specific questions about how to do something a $$$ program can do – using particular programs/products that might be free/cheap and/or that we might already have on hand – might be a particularly attractive use of AI. In other words, I’d love to see more ideas about the judicious use of AI and how we might use the information it provides to help get us get jump-started on some fun projects.
And since the consensus seems to be to keep the thread going… I’m good with that.
Thanks, all.
– David
5 Likes
I had to write a 45 second speech about myself for a nomination video the other day. I tried adlibbing it and one of my guys suggested AI. They ran some on whatever they use, and I tried Microsoft Copilot. Gave it the parameters and some background and it did a way better job than I could have in the time it took. I would have spent hours writing and rewriting.
I have used it to help me with some Arduino code to control my solar panels. I already had a working program but the solution that was presented by ChatGTP was way more elegant, and I LEARNED something in the process.
As long as it can’t make a body for itself…
7 Likes
A little over a year ago I discovered Cursor (Cursor - The AI Code Editor) and it has literally changed my personal and professional life. It’s VSCode with a chat window and the ability to switch between all of the major LLMs at will for one low price.
But it’s so much more than that:
- tab autocomplete is ridiculously good at guessing what you’re going to type
- It indexes your entire codebase and maintains it for context automatically, you can also specifically “@“ files, websites, folders, etc to pull them in as context
- You can save websites as documentation, so for example you can add a GitHub repo or a python package’s “read the docs” link, and now the AI is an expert and will cite the pages where it got its answer from, makes it so easy to learn when you can ask anything and get real answers (most of the time)
- It has “memories” so when it does something wrong and your response is like “don’t use OpenCV it never works, use SciKit instead” it will stop trying to use OpenCV from then on
- Create a Rule file and Cursor makes the LLM reference it at each prompt, so for example in my “cnc” folder I have a rules file that says “The user is adept at programming, science (remote sensing), and DIY but new to CNC.”, “The user has a CNC router with a max cutting area of A x B mm.”, etc. Now its responses will be worded based on my background and fit into my machine’s specifications. Get it set up right and you can get the ultimate brainstorming buddy in Claude 4.
Basically between forcing it to see your context and its ability to iterate on its own work until complete, makes it the ultimate productivity tool. I tell people about it any chance I get 
my teams at work adopted it and we’re soooo much more productive and stress free.
1 Like