Using AI to plan next build

[AleForce1]

As the summer weather continues I have been watching the air traffic kicking into gear on the south side of my house. 10 years ago my wife and I took some farm land next to our house and “retired” it, by creating an acre of land for the pollinators. It’s kind of inspired a few ideas for my next build.

So I used Bing's Co-Pilot to to create a visual guide that I am going to share with my daughter (who is a great artist) to have her paint it .

Here's a few that Co-Pilot made using my request:

 

[Tretgeraet]

The artwork looks really beautiful, I just gotta keep laughing about some of the errors.

I mean, the fret dots are always wrong and then there's that crazy looking bridge on that second yellow telly.

Btw.: If the first yellow had that yellow perimeter like the second one, I'd prefer that.

 

[rockandroller]

That's a wicked cool concept! Pollinators are the mostly unsung Heroes of our planet

 

[NedRyerson]

As an inspirational guide, AI is great. We just have to keep approaching it with caution and skepticism, though I'm also seeing a subset of the population relying on it far too deeply (like job applicants), which is unfortunate.

I recently used ChatGPT to create a unique clip-art style image for a flyer I'm designing for our work picnic. I could've spent hours searching for an anthropomorphic calendar in a life ring in the water flailing its arms with a panicked look on its face ("save the date" -- I was rather proud of myself for how awful that pun was) or spent even longer trying to draw it all out myself and being dissatisfied with the result....or I spent less than 5 minutes with ChatGPT and most of that time was tweaking the initial output to get closer to what I envisioned in my mind.

Just like with playing solos, I have come to grips with my own limitations! My brain apparently just isn't wired that way

I love the monarch design. The hummingbird/butterfly hybrid is interesting and a bit unexpected. Is there some other backstory to that combination?

 

[AleForce1]

As an inspirational guide, AI is great. We just have to keep approaching it with caution and skepticism, though I'm also seeing a subset of the population relying on it far too deeply (like job applicants), which is unfortunate.

I recently used ChatGPT to create a unique clip-art style image for a flyer I'm designing for our work picnic. I could've spent hours searching for an anthropomorphic calendar in a life ring in the water flailing its arms with a panicked look on its face ("save the date" -- I was rather proud of myself for how awful that pun was) or spent even longer trying to draw it all out myself and being dissatisfied with the result....or I spent less than 5 minutes with ChatGPT and most of that time was tweaking the initial output to get closer to what I envisioned in my mind.

Just like with playing solos, I have come to grips with my own limitations! My brain apparently just isn't wired that way

I love the monarch design. The hummingbird/butterfly hybrid is interesting and a bit unexpected. Is there some other backstory to that combination?

I think the previous co-pilot chat keep some monarch in their design. Several hummingbirds have adopted us and I was trying to incorporate hummingbirds into it, but realized that I wanted the body of the guitar to be a metamorphosis of a butterfly wing.

 

[TBurst Std]

I like the 1st one

 

[AleForce1]

When I added hummingbirds to the mix Co-Pilot came up with:


I have a Warmoth Strat body (the one on the left in the photo) that I will probably strip of hardware, paint it black and replace the white pickguard with a black one and then have my daughter finish it.

and make this version:


and then make a Monarch Telecaster. I'd consider doing a VIP version but a Telecaster is more economical.

 

[Rick]

AI is a helper, but, at least in my business when I ask it for stuff it's just plain wrong ... at least for now. It has come a long way, and when it writes something, it's very convincing even when making stuff up.

It would be more interesting to sit down with your daughter and describe to her what you want. Just a thought, why don't you see what she comes up with and then compare it to the AI, and see what you like better. I bet she does a better a job, and it will have a soul.

 

[AleForce1]

AI is a helper, but, at least in my business when I ask it for stuff it's just plain wrong ... at least for now. It has come a long way, and when it writes something, it's very convincing even when making stuff up.

It would be more interesting to sit down with your daughter and describe to her what you want. Just a thought, why don't you see what she comes up with and then compare it to the AI, and see what you like better. I bet she does a better a job, and it will have a soul.

Definitely will let my daughter do come up with the final design and work. It was more a visual exercise to see what AI could do, and it was about 80% of what I wanted.

 

[rockandroller]

I have found that 'not all AI is created equal'.. the most expensive one I use (RUNWAY ML) is hot garbage - complete waste of money with a lot of vaporware promises that never come thru. I find Copilot to be pretty useless, and Gemini to be extremely good (for my needs!). Nothing trained on 'all the information on THE INTERWEBBZZ" is gonna be perfectly reliable (G.I.G.O. !!) but Gemini saves me hours of laborious web-scraping every day just giving me useful hints to help me get some work done faster

 

[NedRyerson]

Definitely will let my daughter do come up with the final design and work. It was more a visual exercise to see what AI could do, and it was about 80% of what I wanted.

It is indeed useful to see if what you're envisioning or hoping for in your mind would actually work visually. Sometimes, an idea should remain just that: an idea and never see the light of day! Other times, what you thought would work ends up working better with a few more adjustments that you might not have been inspired to try without that model sitting in front of you.

AI is a helper, but, at least in my business when I ask it for stuff it's just plain wrong ... at least for now. It has come a long way, and when it writes something, it's very convincing even when making stuff up.

Had that occur to me a few weeks ago. Asked it to write a SQL query on a very, very specific database schema. Since it's a well-known product (at least in the education world), there are plenty of reference materials out there for LLMs to scrape, and the query was mostly right. A few spots were just 100% incorrect, so I started tweaking the model and editing my prompts.....only to realize that the amount of time it took me to teach AI to get even 90% close was more than it would've taken me to just write the damn SQL myself!

Of course, this was just a test to see what AI could do based on me already knowing the outcome and wanting to see if AI could get the answer I already knew.

In this aspect, I think I spend more time prompt-engineering than writing the code myself. I do know in other contexts, the volume of output definitely outweighs the time spent in prompting, though.

 

[AleForce1]

step one finally arrived, a Fralin pickguard with Sunbuckers to replace the white SD Everything pickups and white pickguard.

 

[Spud]

step one finally arrived, a Fralin pickguard with Sunbuckers to replace the white SD Everything pickups and white pickguard.
View attachment 66035

What makes them so sunny do ya suppose?

 

[AleForce1]

What makes them so sunny do ya suppose?

from their site: The Sunbucker Blues is perfect for those who want a traditional humbucking tone, except we magnetize one coil of the Sunbucker Blues stronger than the other coil to give this Humbucker a bright, scooped mid-range tone. This is great for those looking for a brighter, more single-coil sounding humbucker, with a perfect balance of traditional Humbucker tones.

I went with the volume,tone,blender pots and then a Tone Pot Push Pull mod: Phase Reverse Mod.

Looking forward to trying it out this weekend before I paint the body black.

 

[AleForce1]

Kind of like the look of the new pickguard with the body.
The pickups sound great with all of my Synergy modules. Next up painting it black and then dropping if off in KC when I visit my daughter so she can add her design to it.

 

[Spud]

from their site: The Sunbucker Blues is perfect for those who want a traditional humbucking tone, except we magnetize one coil of the Sunbucker Blues stronger than the other coil to give this Humbucker a bright, scooped mid-range tone. This is great for those looking for a brighter, more single-coil sounding humbucker, with a perfect balance of traditional Humbucker tones.

I went with the volume,tone,blender pots and then a Tone Pot Push Pull mod: Phase Reverse Mod.

Looking forward to trying it out this weekend before I paint the body black.

Kinda like more winds on one coil I suppose.

 

[TBurst Std]

It is indeed useful to see if what you're envisioning or hoping for in your mind would actually work visually. Sometimes, an idea should remain just that: an idea and never see the light of day! Other times, what you thought would work ends up working better with a few more adjustments that you might not have been inspired to try without that model sitting in front of you.



Had that occur to me a few weeks ago. Asked it to write a SQL query on a very, very specific database schema. Since it's a well-known product (at least in the education world), there are plenty of reference materials out there for LLMs to scrape, and the query was mostly right. A few spots were just 100% incorrect, so I started tweaking the model and editing my prompts.....only to realize that the amount of time it took me to teach AI to get even 90% close was more than it would've taken me to just write the damn SQL myself!

Of course, this was just a test to see what AI could do based on me already knowing the outcome and wanting to see if AI could get the answer I already knew.

In this aspect, I think I spend more time prompt-engineering than writing the code myself. I do know in other contexts, the volume of output definitely outweighs the time spent in prompting, though.

My experience as well