Everyone remembers their first operating system, right? Well, if they’re a big geek like me, they do. Mine was Workbench 2.0 on the Amiga A500+. Since then, I’ve used various iterations of DOS, Windows, Linux, and macOS – and it got me thinking: what would a futuristic operating system look like?
Basically, what kind of software will we be using on the best laptops of 2033 (and beyond)? If the last few decades of software and hardware evolution are to be believed, the operating systems we’ll be using 10 years from now could be radically different – and yet there are still things we rely on as well (I doubt Microsoft will again abandon the Windows Start menu).
To get an idea of what a future operating system might look like, I decided to ask a few popular chatbots and AI tools to submit designs. After all, who better to come up with ideas for the future of computer interfaces than real computers?
What a great idea, I thought to myself, I’m even going to implement their designs on my own PC to see what they look like. But, dear reader, that was not a good idea. It was a really dumb idea – and now my office is a nightmare.
But whose fault was it? Mine, or the AI I was counting on? Turns out it was a bit of both…
Help with Rainmeter
My first bright idea was to use Rainmeter (opens in a new tab) to create a futuristic theme for Windows 11. Rainmeter is a powerful tool that can radically transform the look of Windows. I thought this would be a great way to quickly implement an AI’s vision of what a futuristic operating system will look like.
I’ve used Rainmeter in the past, and while I’ve been impressed with it, I’ve also found it overly complicated when trying to implement some of the more powerful features. Asking an AI like ChatGPT for help seemed like a great way to overcome some of the obstacles I had encountered before.
You can create interactive desktop elements, called widgets, in Rainmeter using the Lua coding language. Since I had seen some examples of creating ChatGPT code, I thought I would ask it to create widgets and interface elements.
My first prompt was:
Can you create me a cool Windows 11 desktop in Rainmeter?
Unsurprisingly, this request was far too vague, with ChatGPT essentially virtual shrugging and saying, “As an AI language model, I can’t directly create a desktop for you,” and giving me some tips on how to create my own desktop. .
So for my next prompt I went more in depth:
Can you write code for a Rainmeter desktop ini file? I want a widget to give me the date and time, weather in Bath, UK, and widgets that measure CPU, GPU, RAM and hard drives. I want the desktop to be cool and modern, with a sleek taskbar.
Unfortunately, despite the more in-depth prompt, ChatGPT initially replied, “Unfortunately, as a model language, I can’t directly write code or create files.” This surprised me, as I know it had been used to write code before.
Bing, which is based on the same GPT language model as ChatGPT, also started making excuses when asked to create certain things, including code. I feel like this may have been a move by the teams behind these chatbots to limit plagiarism accusations. After all, when an AI generates code, lyrics or art, it takes works previously created by humans.
ChatGPT at least gave me a code for a simple date and time widget:
[Date]
Measure=Time
Format=%A, %B %d, %Y
X=5
Y=5
FontSize=16
FontColor=255,255,255,255
[Time]
Measure=Time
Format=%I:%M %p
X=5
Y=25
FontSize=20
FontColor=255,255,255,255
Rather less helpful, he gave me some general advice on making the other widgets. These were vague enough not to be of any real use.
Still, I put the code in the config.ini file of a new skin, as ChatGPT suggested, and sure enough, a basic time widget appeared. It wasn’t particularly pretty, though.
I was going somewhere, or so I thought. So I tried the “Regenerate Response” option, which keeps the prompt, but allows ChatGPT to come up with something different.
Each time I tried, the answers changed. I tried everyone to see what the widgets looked like. Some worked, some… well, they didn’t. I found myself at one point with a desktop full of ‘%’ icons… and nothing else.
However, it started returning both the code to create functional widgets And instructions on how to add them to Rainmeter.
For example:
[Rainmeter]
Update=1000
[MeasureCPU]
Measure=CPU
UpdateDivider=1
[MeterCPU]
Meter=String
FontSize=12
FontColor=255,255,255
Text="CPU: %1%"
DynamicVariables=1
InlineSetting=1
MeasureName=MeasureCPU
This gave me a working widget that showed me my PC’s CPU usage. I dutifully created each widget and put them on my desktop.
While they were working, they looked very basic, so I told ChatGPT:
These widgets work, but can you make them more modern and dynamic?
He replied, giving me some general advice on how to design modern-looking widgets (not very useful), and another sample code (more useful).
It seems that ChatGPT is currently limited to giving small examples rather than tons of code. ChatGPT told me, “This code uses a round line counter instead of a string counter to show CPU usage, and it adds a percentage text counter in the center of the circle.”
The widget code update turned a rather basic but readable widget into something that made no sense – the round line counter (whatever it is) doesn’t seem ideal for measuring CPU performance. It ended up looking like some kind of demented clock.
My next prompt was change the widget to have a bar chart insteadwhich gave the following code:
[Rainmeter]
Update=1000
[MeasureCPU]
Measure=CPU
UpdateDivider=1
[MeterCPU]
Meter=Bar
MeasureName=MeasureCPU
BarOrientation=Horizontal
X=10
Y=10
W=120
H=20
BarColor=255,255,255,200
BarBorderWidth=1
BarBorderColor=0,0,0,200
BarGradient=1
BarSolid=0
DynamicVariables=1
[MeterCPUPercentage]
Meter=String
MeasureName=MeasureCPU
X=10
Y=35
FontSize=12
FontColor=255,255,255,200
Text="%1%"
DynamicVariables=1
Effectively, this code changed it to a much easier to understand bar graph, with the bar expanding as the CPU was used.
However, it still looked rather ugly and wasn’t the attractive and useful widget I was looking for.
The pros and cons
After spending a good few hours trying to get a working widget, I learned a lot about the potential – and limitations – of ChatGPT and other AI chatbots.
What impressed me was that it produced code that, at a minimum, worked. I also liked how I could fine-tune the code in a conversational style by having ChatGPT change bits to get the desired (or reasonably close) result.
I’m sure with more time I might have gotten closer to what I was looking for. It seemed like an intuitive way to produce code for me, someone with very little coding experience.
However, the limits are also clear. The time it took me working with ChatGPT to produce a very basic widget could have been better spent learning and coding the widget myself. It might not have been so easy, but I would have learned something and probably ended up with something more to my liking.
Another obvious issue was that while the widget was working, it looked awful. Many of his attempts were basically unusable due to the way they were displayed – which made sense, in a way. A computer, after all, knows functionality, but when it comes to design and aesthetics, it has no idea.
This dismal failure is not just ChatGPT’s fault. I should also take some responsibility. Much of the AI’s success depends solely on human input to it, and my prompts, especially in the beginning, were basic and vague – so it’s no surprise that I didn’t get any useful results.
With more time and more carefully constructed prompts, we could have come up with something much better.
The nightmare ends (sort of)
Although the failure of this experiment was disappointing, it gave me some comfort. If ChatGPT had instantly created a perfect OS desktop that looked as good as it worked, I would have been slightly worried.
Instead, it’s clear that ChatGPT and other AIs still depend on human input, and there are things it just can’t do right that humans find quite easy.
We’re not dinosaurs yet, then.
Regarding my desktop, before throwing the whole PC in the trash, I decided to ask an AI art generator to create a futuristic wallpaper. For this, I used Bing’s new Image Creator feature, based on Dall-E 2.
The results were…well, pretty terrible. Completely discouraged, I gave up, picked the least offensive, and set it as my wallpaper. Although it asked for 4K resolution in 16:9 aspect ratio, Bing ignored me and gave me some low resolution art.
After adding the cursed background, I took one last look at my horrible AI-produced desktop, then shut down the PC and unplugged it. I haven’t turned it on since.
#asked #ChatGPT #design #desktop #nightmare
