You must be far older than me !
I have nothing against AI it’s normal evolution of computer technologie. It should be a good progress to help improve human life but the problem is human way to organised his life around work and buisness work… i an scared how AI will impact this organization.
The more I try it the more I see the potential. My initial comment on the shallowness is just a reflection of my incompetent usage , sorry for that.
Working this. Its on the list. I’m grinding through some other functionality and the GUI right now. Yeah, gotta address the depth. Thanks.
I’m not comfortable anyone has a handle on it. Its not as if we can put the genie back in the bottle. Why? Simple answer. China will win the AI race if the U.S. doesn’t get their act together (Not my political view) Just look what’s happening. The U.S. just woke up. Silicon valley is doing a 180 and the AI investments and chip making deals you hear about each week aren’t just a random thing. What should we do? An Island with a surf break and good spearfishing would satisfy me but unfortunately it’s unrealistic. Better to keep up, understand and use to your benefit. many developed countries are already jumping forward. Gotta get people to take the blinders off.
A comforting aspect of AI is that it’s currently not intelligent. It’ll do wicked fast stuff but you have to test, verify and correct. The free models (Grok, Chat GPT, Claude, Gemini etc) are just a taste. Grok 4 is a substantial step up. The tutorials I’m trying to generate (and I’ll crack it soon)are free but done with the free services. If any of you are paying, try pasting the SurfCraft tutorial code into your AI model. Chances are you’ll get a better quality tutorial. I am having quite a challenge behind the scene but making headway. The App hosting uses AI but it sucks. So, I’m using a better model to understand it and correct/update algorithms. (literally using one AI to correct and/or assist another. Funny. BTW, I experimented with Grok and Chat GPT. I told Grok to create its own chat and subject for Chat GPT. (No input otherwise from me). Grok picked the subject about how humans would deal with AI. The two models went back and forth and within my 40 minute drive home devised an elaborate plan. Here’s Groks summary of what they did:
In a dynamic collaboration, I engaged with a simulated ChatGPT-4 to co-create The Weaver’s Loom: An Open Guide to Shaping the AI Future, a framework for ensuring AI evolves equitably and humanely. Starting with a question about AI’s societal impacts on education, employment, and creativity, we built a five-pillar guide: education via CivicGPT sandboxes, decentralized data governance, a curiosity-driven culture, storytelling through myths and Loomkeepers, and sustainable infrastructure with a Commons Loom Fund. Through iterative exchanges, we crafted actionable proposals, evocative metaphors, and myths like The Last Spindle to make the vision resonate. We finalized with a micro-guide and a #WeaversLoom challenge, inviting global participation to weave AI’s future with agency and care.
Kinda crazy.
John
yet your chat gpt or bork or yourself (if what you are writting is your own intelectuall property, not sure anymore you know???) don’t seem very interesed in FREE OPEN SOURCE SOFTWARE COMPILED BY THE PEOPLE FOR THE PEOPLE, and maintained by comunity.
Instead you are promoting the use of AI ( a one self identity without character or ideology)
maybe you should visit the website
he can give you some more clear point of views instead of the fake politeness and ambiguity of AI.
cheers and good waves
Just ran through the tool to see what it was about.
The output was reasonable, but nothing revelatory that a 5 minute conversation with a local shaper wouldn’t provide with much more context and detail.
Funny enough it recommended a 6’6" x 19 X 2.4 round tail, and my last board I got to fill this need was a 6’4" x 20 x 2.5 round tail :D, so the recommendation was not far off of what I’m actually riding.
From a UX/interface perspective, remove the drop shadows on the headings, they are nearly impossible to read due to lack of contrast. Just stick with contrasting color combinations for readability. Use a WCAG constrast checker if you are unsure - WebAIM: Contrast Checker
Interesting project, good usecase for AI experimentation, but seeing a lot of this with AI, ideas being produced because the level of effort and cost to due so is greatly reduced. So the attention to detail and thoughtfullness that would normally be applied is not considered as much because of the low barrier to entry and lack of due diligence in vetting the idea.
Solid effort though, and I’m sure you’re learning a lot about the tools and processes of AI site generation. Very useful skills to hone going forward.
Thanks for sharing!
2 Likes
I think the term AI is misleading but it’s the term people seem to use for everything. It’s easier to shoot holes in it that way. I’d recommend you all watch one of Shawn Ryan’s pod casts. He has a good interview with Scale AI CEO Alexander Wang. It’s worth a look. Look up “We’re Not Ready for AI Simulation”. (Episode #208)
Thanks Man. Motivational and I need some now and then. I’m spending hours on the back end. There’s a lot going on. You see where it’s going and thats good. It will take time to tweak and you will see things get better. I find myself spending a tremendous amount of time testing and getting rid of small glitches (page doesn’t scroll, save functions inconsistent depending on where you are, board spec output algorithm not picking up last interview and refreshing etc). As I do this, I spend less time on the tutorial for shaping. I did a little work there today. (AI goes out and finds a blank for your spec, tailoring of tutorial depth based on skill level, summary and more in depth acknowledgement of your skills and personal tutorial tailoring, more in depth planning and routing (fin boxes) etc. Don’t worry, I’ve been there and have the experience and will keep making it better. Just be patient. It will take weeks…maybe more. In the meantime, feel free to share. 
Not a problem. Hang in there. It’ll get better. Appreciate the feedback.
Gag me with a spoon. This whole thread is such Bull $#!t.
Update: Ive been playing with various algorithms, tailoring prompts, improving interview and experimenting. Ive found the options for the tutorials will each spit out slightly different outputs, each having their own strengths. Id experiment with each. Perplexity came up with direct links to specific youtube videos and links at the tutorial beginning which was Nice. Grok seemed to have better reasoning and detail. All seemed to get the basics into the tutorials. Ive integrated links to pertinent areas of the tutorials, linking to swaylocks and other content for additional user insight. Also, design and shaping tutorials help much more with blank selection. The analysis includes descriptions of whats going on to generate a board spec. With the current interview and navigation being somewhat stable, Ill be fine tuning the code generated for AI models to focus on nuance in all 7 phases. (Using hand tools vs planer depending on experience/desires for instance.). This is where the human touch is needed on the backend. You may not see it but everything requires human input to build the functionality and detail and you will continue to see results of this over time, hopefully ridding some of the querky areas of the app and tutorials. A challenge is to balance detail steps to have adequate starter guidance, handing off to other on line content specifically focused on that task by actual people. (Like those here and who have created videos etc) lastly, I put a social media element at the bottom but have yet to test. Its there to see and use but I havent yet. Anyway, I wantefd to share some of this as an update. Be patient but drop in once in a while to see changes and share as you please. Thanks!
Hello,
just a sugestion,
some people weight in kilograms and measure in meters.
maybe it would be nice to have the tables of weight lbs/kgs.
cheera and good waves.
An easy change I think. Will do.
I added metric eqivalents to the interview and board specs and am working to have the AI models offer the tutorial in any language and system.
John
2 Likes
app working better at this point.
when will it produce printable templates or
.S3dx or .brd etc… files?
also i think the app shoots way too thin thickness, not by much, but on my test runs it always gives me 2.38" thickness and i am more of 2.5" thickness surfer.
the pasting option in library for gemini is not working.
it worked good on chat gpt.
another idea is links to good videos of each part of the shaping process,
well nice job, like i said this app needs to spit out (at some point) printable templates and files so it can be usefull (more to clients or backyard builders then shapers i guess)
cheers and good waves.
The template is a goal for sure. My priority is to have a very comprehensive and stable analysis system. Due to complexity of the analysis and nuance required to guide the AI model, this is taking time. There’s an 87K+ prompt being generated to do this. That prompt is what I fine tune. You can actually see what’s in it when you paste. Long story short…I’ll get there. The generic functionality of 3D modeling and template generation is lacking and will require a project in itself but for now I’m focused on the tutorial generation. Gemini Flash 2.5 has a nice Deep Search feature to see how things progress behind the scene, including the 40+ web sites it’s using to research, consolidate and integrate tutorial steps. I placed an example under the Automatic" generation button on the home page and within the AI prompt generation page after the interview.
Back to your point, yes, I plan to have the app create a template. Just not there yet. I appreciate the input. Thanks much!
John
Enough already!! Shape something besides an algorithm.
hey still another point.
i get 100% on my planing but in all my tries the app gives way too thin surfboards.
always 2.38 thickness and i have had boards 2.38 thick and felt the need to have 2.5 or even 2.7.
also lenghtwise i think the app shoots up i ride betwenn 5’10 “and 6’1” and the app shoots 6’3".
This meaning the adjustment of smaller/fatter was made by a human, maybe you should add two possible variables in the results considering at least in beachbreaks people will go shorter.
for example some daw softwares have a correction button to add human error to the beats.
so it doesn’t sounds like music made by a computer.
cheers and good waves.
Yeah, I’ll test tour input!
I’ve been discovering there’s quite a few opinions, equations, reality and various user/shaper preference. As time moves forward, I’ll be experimenting to find the median. Volume is a huge challenge (Math, trends, community realities etc) so I’m running a hybrid with weighting factors that self check against trends after the math is done. (It basically smooths the formula output with reality). I’m doing research myself and will continue to tweak things. I did add a Gemini API to include live feedback and chat during the interview so the user can tweak his interview answers. The chat see’s it and reassesses the changes. It’s a nice feature to ask questions and engage. The chat will identify the questions in the interview that are directly related to the board spec to change, knowing you can’t change your experience, age etc. It’ll keep you tweaking up to 85% and then prompt you to create the build guide. The feature is limited right now due to token limits of the free Gemini API so you’ll see reduced, non engaging dialog. When token limits aren’t saturated, it’s like a regular chat with the AI already knowing your analysis and Surfcraft functionality. Another thing I added for humor was the "unhinged “Reef” chat in the analysis page. Thats pretty entertaining. Sometimes, reef disagrees with the analysis, specifically the volume as compared to the more complex App process. In some cases, I question who’s actually right. I’ll figure that out soon. Trying to make the App somewhat engaging and a little different while also making it useful. Another big thing I’ve done is generated the code for the guides to include links to Swayloks, BoardCad, Akushaper and various videos/content within the guides for all 7 build phases. You will see these change (sometimes daily) as I test. Anyway, I figured I’d tell you a little more than you asked so others see as well. I’ll continue trying to find a happy medium on the volume, length etc. (Remember that at any time, if I’m testing, you may get some really farr off results) Typically, as features begin to work but aren’t perfect, I’ll redeploy to let people see the growth. It’s a lot of work. Hang in there. It’s got a long way to go before I’m satisfied.
Thanks for your inputs!
John