[TriEmbed] The enshittification of Arduino begins? Qualcomm starts clamping down @itsfoss2

Sun Nov 23 12:52:18 CST 2025

I think I remember you are doing this for work, right? In which context
maybe a quick end result is worth more than personal enjoyment of the
process and personal skill development? Do you find the process enjoyable?
Most people don't think of being a project manager as a fulfilling activity
worthy of hobby time.

To me the beauty of Arduino and other libraries is that they give you the
personal satisfaction of figuring out how to do what you want. Granted a
library is usually doing hard work that you don't understand. But there is
a consistent abstraction layer and you can gain expertise with using it. In
really weird situations you can even dig into the open-source library code
for the details. (I too have used Paul's Teensy Audio library and on a few
occasions had to dig into the code. It was also well beyond my ability to
write, but I was able to understand it well enough to figure out my
problem.)

I think I mentioned my rowing machine project on one of the meetings. I
could easily buy a new computer for it. Or I could use a full-featured
open-source rowing-machine project. I find it much more satisfying to make
my own "from scratch" (though using basic libraries to do a lot of
heavy-lifting) even if it is feature-poor compared to the other options. If
I had to churn one out for work I would make a completely different
calculation about that.

To me GenAI is the ultimate enshittification of the entire hobby tech
industry. Want to debug your AI-generated code? Have to pay a big tech
company for more GenAI tokens to analyze it for you, because you don't (and
possibly can't) understand how it works. We don't currently pay the real
cost either, since everything is subsidized by investment dollars to
generate more growth. It will be interesting to see if the cost comes down
enough to offset the necessary end of that subsidy.

When I was a kid, I could make something on the same order of magnitude (in
functionality) as the commercially-available software, writing every line
of code. By the time my kids came around, that wasn't really possible for
PC/console (and later phone) software. However it was still roughly true in
the simpler embedded space, with the help of libraries. (Maybe I have
simplistic view of this though). With GenAI, I wonder if there is even any
point in kids making anything. Just ask the GenAI directly to do the thing
you want instead of spending effort in writing software (even with GenAI
help) to do it.

I have used GenAI, for example to create a logo for my bowling league. My
skills were utterly unmatched with the quality I wanted, and it's not a
skill I want to spend time learning. I didn't feel any enjoyment in in the
process but I was satisfied with the result (after some additional manual
tweaks). And I use it when coding at work. But it seems sad to me to have
it used everywhere in places where people could instead be training their
own internal LLMs by doing things more directly and making and correcting
their own mistakes.

Interested to know what other people think.

On Sat, Nov 22, 2025, 06:47 Trampas Stern via TriEmbed <
triembed at triembed.org> wrote:

> In my personal opinion Arudino has been dying a slow death for years,
> mainly from then number of bugs in their code and libraries.  However it is
> really falling fast now because AI will be taking over that market area.
>
>  In the last ~6 months AI has gotten really good at firmware/software
> development.  I had AI write an entire web app that monitors remote servers
> data logs and alerts me on failures, I did not write or edit a single line
> of code. Around 80%-90% of code I develop is AI based now (even embedded),
> as such I no longer care about the programming language, APIs, etc.
>
> For example, I now write chip drivers by feeding datasheet to AI where it
> writes the driver for me, and will not go back to doing it manually ever
> again.
> Needed FATFS and SD card driver, just told AI to add to project and
> include test cases to verify it worked.  Which it did, and worked on the
> first try, of course this was Zephyr based project so not a huge lift, but
> it worked.
> Even Zephyr device trees are insignificant to AI, I don't have to figure
> out convoluted syntax or setup, I just tell AI what I want.  Even had AI
> write python test cases to run on desktop which verify that Bluetooth
> worked correctly.
>
> The point is that when you can ask AI to setup a development container and
> IDE for your embedded project, then write code and test cases, so there is
> little need for Arduino.
>
> I should mention that I was debating on learning Rust for embedded
> development last year, I choose not to.  Specifically, AI is at the point
> that the programming language is irrelevant. That is AI will be able to
> translate code from one language to another with little effort. Also it
> will be able to verify that C/C++ is just as safe as Rust code.  As such
> embedded development will be moving to a job of project management, where
> you are managing the project requirements, testing and tasking AI with
> writing code.
>
> Even on the webserver project AI was having a hard time with javascript
> component, kept having lots of syntax mistakes. So I told it to write a
> lint tool to verify its results. So it wrote a javascript lint tool and
> would run it and fix its own mistakes.  Basically AI today is as good as
> college level intern.  That is it makes some dumb mistakes but can be
> highly productive with the correct "management".
>
> I see  AI technology is much like computers were in the 80s that is
> everyone could see how they would be use and the advantages of having one
> on everyone's desk.  It will redefine the job market much like computers
> changed the typing pools at large companies.  That is there will be a shift
> in skills you need for a job towards AI.
>
> The point is if you are worried about Arudino, instead take some time and
> try the latest AI using copilot and VSCode, you might find like me that
> Arudino is insignificant, like a typewriter.
>
> Trampas
>
> On Sat, Nov 22, 2025 at 1:13 AM Dwight Morgan via TriEmbed <
> triembed at triembed.org> wrote:
>
>> The enshittification of Arduino begins? Qualcomm starts clamping down
>> @itsfoss2
>>
>> https://share.google/03WWYzZ9EJkguCpF6
>>
>> Sent from my iPhone
>> _______________________________________________
>> Triangle, NC Embedded Interest Group mailing list
>>
>> To post message: TriEmbed at triembed.org
>> List info:
>> http://mail.triembed.org/mailman/listinfo/triembed_triembed.org
>> TriEmbed web site: https://TriEmbed.org
>> To unsubscribe, click link and send a blank message: mailto:
>> unsubscribe-TriEmbed at bitser.net?subject=unsubscribe
>> Searchable email archive available at
>> https://www.mail-archive.com/triembed@triembed.org/
>>
>> _______________________________________________
> Triangle, NC Embedded Interest Group mailing list
>
> To post message: TriEmbed at triembed.org
> List info: http://mail.triembed.org/mailman/listinfo/triembed_triembed.org
> TriEmbed web site: https://TriEmbed.org
> To unsubscribe, click link and send a blank message: mailto:
> unsubscribe-TriEmbed at bitser.net?subject=unsubscribe
> Searchable email archive available at
> https://www.mail-archive.com/triembed@triembed.org/
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.triembed.org/pipermail/triembed_triembed.org/attachments/20251123/f8c6ef5d/attachment.htm>