Android Question What is the best AI for generating B4X apps?
- Thread starter rtek1000
- Start date
- Similar Threads Similar Threads
Look, it is incredibly useful to get an idea moving towards a program solution from AI. However, there is still a big difference between generating lines of code and developing real, well-tested error-free programs.
What I don't hear anyone talking about is that, ultimately, it all comes down to money. Right now, you can ask all sorts of questions "for free," and you can give away your program source code to train their AI model. Once the AI model is trained well enough, you get to start paying.
What no one is asking themselves out loud either is how a programmer in the future will distinguish themselves from an AI model trained by their profession, and how to prevent a competitor from using your intellectual property. Musicians are already speaking out, because music plagiarism is easier to spot than source code reuse by third parties.
Those trying to earn a living with this program face devastation, because why would a client pay you for work that you have now given away for free while training an AI environment?
Upvote
0
I utilise Claude AI, it's been a game-changer. The projects feature is super handy - lets me collate all related chats and docs in one spot, so I don't have to give context every time I start a new chat. I've also used Claude to whip up a little b4j app that consolidates all the .bas files for an app into a single text file, which I upload to keep track of updates (since .bas files aren't supported).
I've got 4 b4a apps syncing with MySQL on a VPS, plus a b4j dashboard app for BI reporting, and a b4j server app managing MySQL sync via jRDC. All these apps are linked in Claude's projects section, making it a breeze to update and debug the whole setup.
I've got 4 b4a apps syncing with MySQL on a VPS, plus a b4j dashboard app for BI reporting, and a b4j server app managing MySQL sync via jRDC. All these apps are linked in Claude's projects section, making it a breeze to update and debug the whole setup.
Upvote
0
I used to employ 23 people making, selling and supporting software.
I now work alone and, with the help of AI, I can develop an App in a month or two, which previously took at least a year or two.
I am not embaressed about this, having cut my teeth on MACHINE LANGUAGE Programming (00101100).
You have no idea what it was like to suddenly be able to program in Cobol and Basic
Today we jump from Coding to Visualising what we want - Sorry ... you don't have to write even 1 line of code anymore. (This will definitely disappoint a few)
KEY If you can visualise it - it can be done - its like having a team of 50 programmers working for you who understand your every thought.
As an example of what AI can do, this is what I keyed into ChatGPT, and the Attached file is what it gave me in about 2 minutes. (This is about image creation)
This can also be used for developing complex applications
Those of you who have experience in AI will probably laugh at how I have used ChatGPT
THIS IS WHAT I KEYED INTO THE CHATGPT PROMPT
I have a 12m x 6m "hothouse". It is currently a frame - very strong and very heavy. It does not have any covering at this stage. I want to make it a "hothouse" being able to grow various vegetables and herbs and fruits in Australia, both in and out of traditional seasonality. I also want to reserve a section of the "hothouse" to be a entertainment area, about 3 m x 6m wide (this is where you walk into when coming into the hot House, with electrical appliances - eg, a bar fridge, coffee making facilities etc along one side and on the other side an area I will call a Chinese Writing room (about 2m x 2m" setup to look like an old fashioned chinese writing room with table, decorated walls, and an antique looking table and chair. Between the writing room and electrical appliance side will be a square table for guests to sit around. there will be a BBQ somewhere also. The rest of the "hothouse" will be for growing plants but will have a walking corridor so that people can walk around, and ride mobility scooters. Looking for some ideas on how to make the "hothouse" a warm and inviting area for all who want to come and enjoy hospitality.
KEY - make sure you tell AI exactly what you want (if possible)
I then press ENTER
The attached file is generated and displayed on the screen, and finally I ask CHATGPT
I would love a sketch and if possible some images of how it would look internally, especially the hospitality area with Chinese writing room.
I then press ENTER
That's it (a kid could do this)
See the attachment for the advice offered and finally the PICTURE of what it might look like (Its at the end of the Word.docx document)
Attachments
Upvote
0
There is an important thing a child does not stop to think about, and that is intellectual property. But a child also does not like it when another child uses their drawing and presents it as their own work.
Since 2007, IBM has been developing the supercomputer Watson. By having Watson read a large number of reports very quickly, it was able to produce an extremely accurate summary. If you do that in collaboration with the source provider, there is nothing wrong with it. But what you are seeing now is that writers, musicians, and authors are starting to voice their concerns regarding intellectual property rights.
One of the conditions for using open source is citing the source. The creators of open-source software complain that AI omits this for results that look very similar to their source code. This raises the question: when a software supplier is accused of plagiarism, how can they defend themselves? They cannot, and that can have costly consequences.
The first users have an initial head start, but are quickly overtaken by others offering the same solution. So, after some time, there is nothing left to distinguish your program from the competition. This leads to low prices and competing on services provided.
Open source solutions will decrease, thereby slowing down the discovery of "new AI solutions." The best opportunities lie with software developers who develop their own ideas, who may well be inspired by "AI suggestions" (which currently surprisingly often do not always work immediately: how do you solve that challenge without programming knowledge?), and who understand how the program works in order to distinguish themselves from "one-fit-all" AI solutions and refute allegations of plagiarism.
Upvote
0
What I usually do is find out from the LLLMArena Leaderboard in terms of which best voted coding agent.
So im testing the ones that Ollama has and currently using glm, gemma, qwen-coder, deepseek and I change inbetween.
So im testing the ones that Ollama has and currently using glm, gemma, qwen-coder, deepseek and I change inbetween.
Upvote
0
I have to disagree with the comparison. Creating an image is fundamentally different from generating code. An image is subjective; whether it has 'darker greens' or 'bigger plants' doesn't make it 'wrong'—it just makes it a different variation.
Code generation doesn't work that way. It's deterministic and logical. If you master the language, AI is an incredible multiplier that helps you code much faster because you can spot errors instantly. But if you use it to program in a language you don't know, you're building on sand.
AI often generates code that looks correct and works for the 'happy path,' but it fails miserably in edge cases. This becomes a nightmare in a real-world scenario. If you build an app for a client without knowing how it actually works under the hood, what happens when a bug arises or they ask for a custom modification?
You are forced to ask the AI to fix it, which often triggers an endless 'error loop.' The AI tries to patch the issue, breaks another part of the logic, you feed the new error back to it, and it gives you another broken patch. Without understanding the code to break the cycle, the app quickly devolves into an unmaintainable mess.
Don't get me wrong, I'm not saying it's impossible; But this approach is usually better suited for prototypes, MVPs, or demos. Moving from a demo to a robust, production-ready app is a path full of hurdles if you don't have a solid grasp of the language.
Last edited:
Upvote
0
