La Artificial Intelligence It's something that, if you label it as a "fad," you can't deny that it's not temporary or short-lived. We've been talking about it for at least three years, and in reality Don't go beyond Machine Learning and reach the next level. This doesn't diminish the hype in the slightest. As is the trend, the various media outlets don't miss any opportunity to publish something new on this topic, as this article shows.
The fact is that recently there are quite a few blog authors who claim that They are using fewer and fewer applications and relying more on chatbots. of Artificial Intelligence as Gemini, ChatGPT, or Grok. And well, they're partly right. Don't be offended if the editor who said they used a calculator AI reads this, but it seems a bit excessive to me: you can do calculations directly with the voice of Siri or Google Assistant without even unlocking the screen. Unless he meant this very thing, in which case I agree.But we have to be honest and say that they don't do everything well, nor do they come close.
Image editing yes, scaling no. At least with AI in Chat.
It was the last article about replacing apps with AI, I don't remember which one, that made me think, "I'll try rescaling images." I usually use Upscayl, my preference over Upscaler Because it looks better in KDE. Sometimes I find that a project publishes a release note with an image that's about 600px wide, and currently, on the Actualidad Blog network, we use 1200px images. If I upload the original image, WordPress will simply enlarge it, and it will look somewhat blurry.
So what I do is use a program that increases the size with little loss of quality and I usually get decent results.
The problem is that it's an application that uses Electron and I have to use its AppImage, which I don't like, or install the AUR version, which also installs that Electron that I don't want. It was these pebbles in my shoe that encouraged me to try Gemini, which is now in the news for its Nano Banana.
What does Gemini Nano Banana do?
Nano Banana can deliver impressive results, but the conversations to get the edit we're looking for are sometimes excruciating. It often gets it right the first time, but often it doesn't, and asking it to edit the first thing we came up with... sometimes seems more like Artificial Folly than Artificial Intelligence.
In my test, which I also used in ChatGPT and Grok, I took a screenshot of my desktop, reduced it to 400px wide, and asked if it could resize. When he said yes, I told him he'd have to analyze the image and try to return it the same size, but with a minimum width of 1200px. The funniest thing is that he thought it would be a good idea to add the Windows trash can icon to the top left corner, but what wasn't so funny is that he didn't understand what I was asking: to increase the size without losing quality. What he did was increase its size like we can do with Gwenview, leaving everything blurry (and with that trash can).
I got the same result with Grok and ChatGPT, and at that moment I started to think about other Artificial Intelligence models such as Easy DiffusionThat was just before I thought about it and realized what a stupid idea I was doing: if I'm not going to use a chatbot and have to find an alternative, why don't I use the same one I use now (Upscayl)?
Artificial Intelligence (Machine Learning, ahem) will continue to improve
So yes, we can forget about many apps and do the same with AI chatbots, but they must continue to improve to cover more tasks with confidence. In the medium term, we'll be doing many tasks with them, but it's still a bit farther before we forget about what ALREADY works.