ich denke das Quelltext aus einer KI irgendwann einer Nutzungs Lizenz unterliegt
und damit geschuetzt ist. Offenlegung es Quelltextes wenn man KI benutzt hat und Abmahnwelle …
Rechtsanwälte wittern Beute.
1 Like
i see that we humans make things over complicated and then we need/use a AI for helping out of trouble.
1 Like
This is a long and boring thread about a non-issue needing a “do nothing”.
So, to summarize:
-
AI is here and even with errors it won’t vanish, it will only evolve.
-
People are using it frenetically, for anything, and it’s evolving and will enter our lives one way or the other even if you don’t notice or don’t like it.
-
As it will get into our way as watching TV, or preparing a mug of coffee, it does not deserve special labels, with few exceptions to be identified by authorities. If your software would be entirely written by an AI, it should, for example. There’s a high risk involved in this case and users must be warned, and you should be blamed for the results of doing it. As a tool, used as a helper for small supervised tasks and audited content, it is like someone doing a copy/paste from a snippet from this forum for example.
-
For software, as helpers, there’s no return, they are already here, and everybody will have the tools to help or to make a mess, but they will have it, one day or another, even if that is a huge plugin with a huge special trained model with it.
-
No, I’m not using them for coding, not yet, but yes, I do lots of experiments and intend to use them for small tasks when possible. And yes I already used them to things like imagery and even trying to refine some algorithm with little success and many errors. And yes, I intend to use them in the future as helpers when they become good enough; no one will have much choice, they will be around.
That said:
-
So, no. No need of special categories. You will ask for things about code here, not about AI. If your question touches AI, because whatever, the general category or off topic will serve you.
-
Human Authored Certification is a fallacy, it should demand an inspector accompanying the entire development of something or the certification is basically a lie to grab someone’s money.
3 Likes
There is no skill loss. You have to be even more knowledgeable when using AI. I highly doubt that AI will be able to reason soon.
What is the difference between the successful and the non-successful AI users here? I’ve been using AI now for almost 2 years for:
- php scripts
- CSS
- SQL
- AppleScript
- Excel
- syntax errors
- regex
- preparing testimonials
- translations
- speed improvements
- competitor analysis
- blog posts
and so much more.
4 Likes
I would say that, probably more than anything, I’d rather just write the code myself than spend a lot of time iterating over specific instructions to an LLM just to get a result that I then have to code review anyway, and in some cases end up writing myself from scratch even after all of that.
For research, like asking questions about POP versus SMTP, I agree that it can be useful. Likewise, I also use it for copy editing on my sites and marketing materials. Those use cases, in my experience, take far less hand-holding.
For things like PHP, there’s a massive amount of data that it’s been “trained on” and has stored for answering such queries. Which brings me back to what I previously said:
4 Likes
(Bing Translation:) I think that source code from an AI will be subject to a license at some point and thus protected. Disclosure of the source code if you have used AI and wave of warnings
Well, never say never.
There have been patent infringement cases that could be a precedent, like when Oracle (after buying Sun) sued Google after discovering copied Java code in Android.
Then there was that whole mess about ownership of the JPEG file format and the dubious collection of license fees from some customers.
1 Like
Well, I had to use ChatGPT to find it. 
The Xojo Forum operates on the Discourse platform, which offers features to manage your notification preferences for specific categories. To mute a category and minimize notifications from it, follow these steps:
- Navigate to the Category :
• On the forum’s homepage, click on “Categories” to view the list of available categories.
• Select the category you wish to mute.
- Adjust Notification Settings :
• Once you’re in the desired category, locate the bell icon near the top-right corner of the page.
• Click on this bell icon to reveal a dropdown menu with various notification options.
• Choose the “Muted” option from the list.
By setting a category to “Muted,” you will no longer receive notifications or see new posts from that category in your feed. This allows you to tailor your forum experience to focus on the topics that interest you most.
Of course, as with all things AI, this might actually not be the best or right answer. But it is a start.
2 Likes
That works.
Though, there is one scenario worth considering to support trying to separate the AI-related posts more consicely.
Note: this scenario assumes AI bots are reading the Xojo forums, of course.
This scenario has to do with folks posting code here generated by AI (knowing it’s wrong) and then asking for help about it.
That wrong/broken code (potentially containing methods or language features that don’t exist in Xojo) is potentially polluting the quality of the source that AI is consuming about Xojo - furthering the inaccuracy of future AI queries about Xojo.
To fix that, it would help if all the potentially bad code was kept in one dedicated forum category - then try to the prevent scanning of that category (robots.txt?)
Technically, it may not be possible, but I thought it worth mentioning.
Thank you for your feedback Rick
1 Like
If you want to mute a sub-category, such as some posts in a language you don’t speak, find the sub-category (in International, in this case) at step 1.1 in your steps above, and click it. Then proceed from step 2 above.
2 Likes
AI must filter out things in context. A “This code is broken, help me” from some guy, made or not by AI should be considered not safe at any context. Maybe they try to focus on things marked as solution, or “loved”, and not said as broken/wrong in the context.
A forum demands much effort to be a source of code. Not sure if they will trust it. They investigate source repositories as GIT for that.
You’re welcome.
1 Like
Bingo! The way I’ve explained it to people is that it’s similar to giving a scalpel to a surgeon. AI is just a tool but one needs to be competent/expert enough to know how to use it properly. If you give the scalpel to a layperson and send them into surgery, then bad things are gong to happen.
As you also point out @Anthony_G_Cyphers, AI is not a replacement for knowing Xojo, its frameworks, the IDE, etc. But if you have at least a base level of competency with Xojo, then AI in my view can help accelerate your efforts.
1 Like
I get the resistance to AI. I really don’t like it, I find the trend of adding it absolutely everything to be pretty irritating. I try not to even call it AI since it’s not actually, well, intelligent.
That being said, it has a place and outright dismissing it isn’t helpful. There are people for whom it’s a very useful tool. Like me.
The key is knowing how to use it, which I’m slowly beginning to grasp. Case(s) in point:
We have a reel to reel tape deck that was modified in the late 1970s to play a special kind of audio tape that looked like (same dimensions and sprocket holes) Super 8 film. But it’s audio tape. This had an outboard piece of hardware that could “resolve” fluctuations in speed to allow you to play back audio at a fixed speed so it stays in sync with separate picture. Trust me, it works. It’s how the soundtrack on every movie you watched until digital audio editing, was made. But we don’t have that extra hardware, just the modded deck, and it was so rare in this format that you can’t just go out and buy the part. So with the help of a local engineer who was familiar with the deck, we figured out what we needed to do to feed back into it to control its speed. I built a simple resolver with an Arduino, but it can’t output a high enough voltage. I needed to use an op-amp to boost the voltage. I asked on the usual forums online and got 20 different answers, with no real assistance on how to do it or explanation of how it works. Threads turned into arguments between the people who were supposedly helping (not to mention all the people who told me I shouldn’t bother building what I’m building in the first place because “it will never work” - despite 60 years of industry use of this very setup). So I asked ChatGPT and even though it took a little back and forth, it all works now and it was a pretty painless process.
Second case: I’m working on something in Xojo and need to draw a line on a canvas. In all the years I’ve used Xojo, I’ve never had to do this. I spent 2 hours in the documentation trying to figure it out and just couldn’t. I’m not a programmer by trade, but I do write apps for in-house use at work. I don’t do this day in and day out, so I don’t always know the right way to ask something, nor do I necessarily know the “right” way to code it. I’ve asked for help (here and elsewhere) many times, and while I’ve gotten some really good help here, I’ve gotten just as much smug attitude and responses like “read the manual,” which isn’t helpful. ChatGPT had the code up and running in about 20 minutes.
FWIW, I didn’t bother asking how to do this here because the last time I did, when I had this idea a few years ago, I was told that it infringes on a patent and I shouldn’t do this. It’s an in-house app that’s not for sale. The company who holds the patent doesn’t care. I don’t care. I never got an answer back then, but the project was back-burnered for other reasons.
Anyway, in both cases ChatGPT made mistakes. With the circuit design, I asked some EEs I know to verify the schematic, which they all said was correct. Except it didn’t work. None of the humans who looked at it noticed that the pinout on the opamp I chose to use was ever so slightly different than the one ChatGPT designed around. When I asked it what part its circuit was modeled on, and then told it to redesign for mine, it worked instantly. Took me a few days to figure that out, but it did it with no attitude.
With the Xojo code, I recognized some mistakes immediately (using .invalidate on a canvas paint event, or “+=” notation). When I made it aware of these things, it wrote valid code.
I’m not saying that there’s no place for humans in this kind of interaction. Far from it. But you will get a non-judgemental response without the smug superiority you get from (some) people. And having worked for and with tech companies for almost 30 years on hardware and software development, having done Software QA for a living for several years and dealing with an ego-bruised programmer who tried to get me fired for doing my job (finding bugs that happened to be in his code), I can confidently say that it’s a breath of fresh air to ask a question and just get an answer, without the extra sprinkling of attitude. Is it always right? no, it’s usually wrong the first few times. But as a human, who has a reasonable ability to troubleshoot, I’m finding that I can coax a good answer out of it fairly quickly. It’s becoming an invaluable tool for getting me on track quickly.
Can ChatGPT write you a Xojo app out of the box? No, of course not. You need to know something about the language already, and know how to frame the prompt, and how to limit its scope. Once you do that, you get pretty decent results. Certainly results that get me up and running and that I can build off of, without getting into a religious war on a forum over the best way to do something.
BTW - would I like to see ChatGPT trained on Xojo and integrated into the IDE? 100% yes, please! This is absolutely an area where I want to see more AI, not less.
3 Likes
That inspires me. Maybe on my site I shall boast, “100% AI-free code. Cage-free is another matter.” 
4 Likes
When I see code on the forum, I sense resistance to the use of the modern Var instead of the older Dim. These are entirely interchangeable. Whatever the reasons that people have for sticking with Dim, it can be helpful in trying to distinguish code from the newer versions of Xojo from older ones if Var is used. So I urge people to use Var when posting. I wonder if this understanding will works its way into AI. One of the difficulties that AI has in its understanding of Xojo code is its use of deprecated or even now absent commands in its suggestions. Not to mention, of course, its injections of code from other languages.
2 Likes
For those that are interested, a short article that summarizes a new study’s findings.
Microsoft Study Finds Relying on AI Kills Your Critical Thinking Skills
Some excerpts: “Researchers from Microsoft and Carnegie Mellon University warn that the more you use AI, the more your cognitive abilities deteriorate.”
“…The study does not dispute the idea that there are situations in which AI tools may improve efficiency, but it does raise warning flags about the cost of that. By leaning on AI, workers start to lose the muscle memory they’ve developed from completing certain tasks on their own. They start outsourcing not just the work itself, but their critical engagement with it, assuming that the machine has it handled. So if you’re worried about getting replaced by AI and you’re using it uncritically for your work, you just might create a self-fulfilling prophecy.”
4 Likes
It wasn’t long ago that a fellow disputed me, on a topic on which I am well-versed. I sensed that he had Googled it, and did so myself. What he had done was to literally copy/paste the Google AI summary. It was clear that he did not understand that, either.
3 Likes
This has become far too common. I’ve been seeing a lot of
All you need to do is X!
When pressed, I find out that the person had asked AI and gotten not only an incorrect answer, but took it as gospel then acted as if I were clueless.
3 Likes
I will take you at your word, because I shall never get anywhere near X.
1 Like
X meaning <some thing>. Not the social media platform or an illicit substance.
3 Likes