in

Ghibli -style images have firing the chatgpt phenomenon again

You probably remember that moment in the late 2022 in which Chatgpt broke into the scene. An application created by a practically unknown startup allowed, for the first time, to maintain a fluid conversation with an AI. We could ask him to write a story, review a text or explain string theory.

At a time when the industry seemed to advance with incremental improvements, without great surprises, Chatgpt’s irruption was an unexpected turn. And yes, he caught everyone with the Guardian. Because, as happens so many times, when a technology catches us, we run in mass towards it. The chatgpt phenomenon was no exception.

He only needed five days to reach one million users. A shocking figure if we compare it with other technological giants: Netflix took three and a half yearsTwitter two years, Facebook ten months and Instagram two and a half months. Now, Openai is news again: he has done it again, overcoming his own popularity records.


Publication Altman Chatgpt
Publication Altman Chatgpt

Click to see the original message in x

The person in charge of announcing the new milestone was Sam Altman himself. “The launch of Chatgpt 26 months ago was one of the most crazy viral moments I have ever seen, and we added a million users in five days. We just added one million in the last hour,” The OpenAi CEO wrote in a message posted on X on March 31.

There are very few services capable of receiving such an avalanche of users in such a short time. One of the closest cases starred in Goal with Threads, Your alternative to Twitter, which added two million users in just two hours. Of course, the context was different: direct integration with Instagram greatly facilitated this mission.

But not everything stays in the official data. Analysis signatures such as Tower Sensor also reflect the huge growth of Chatgpt. According to your estimatesapplication downloads increased by 11% last week, while the number of active users grew by 5%, with subscriptions by increasing 6% in the same period.

When technology connects with the public

It is no secret that artificial intelligence has evolved at high speed in the last two years. However, they are not always the most sophisticated technical advances – Those that seek greater performance, solve complex problems or democratize access to reasoning models – those that manage to connect with the general public.

Generally, what moves us is on another plane: nostalgia, art, humor. Or, directly, the memes. And it is precisely where the last of Openai has found his hole in everyday life. Exactly seven days ago, Chatgpt was updated With a new function: an integrated image generator, driven by the multimodal model GPT-4O.

Pictures Chatgpt Studio Ghibli 1
Pictures Chatgpt Studio Ghibli 1

One of the many scenes recreated with chatgpt

Although GPT-4O was already present in ChatgPT-we met last year-until now he stood out for his skills in text generation and computer vision. But It did not generate images. That has changed with this latest update, which, curiously, is especially good when recreating images in very recognizable styles.

In Spain, as in other parts of the world, users soon took advantage of the novelty to unleash creativity. They began to recreate iconic images In the purest Studio Ghibli style, transform vacations in scenes built with Lego pieces or convert real portraits into versions of Muppets or in detailed illustrations Pixel art.

But not everything is so idyllic. Fever for the new ability to generate images in ChatgPT has been accompanied by two factors that, although less visible, are not minor. The first has to do with The infrastructure: demand has been so high that OpenAI delayed deployment for free users and impulse use limits.

In addition, the question of visual styles is not exempt from controversy. As users experience with recreations inspired by very recognizable creative universes, a debate has been on the table that has been around the development of artificial intelligence: how have these models learn?

The answer, although not always transparent, points in an awkward direction. Many of these models, including the one that drives this function in chatgpt, They have trained with large volumes of images available on the network, Many of them protected by copyright. This, once, is a point of tension between technological and authors.

Images | Xataka with chatgpt | @MDURBAR | X screen capture

In Xataka | How to turn your photo into an action doll with accessories using chatgpt

In Xataka | Openai has just lifted the greatest financing round in history: there is a blind faith in the AI ​​despite everything

What do you think?

Leave a Reply

Your email address will not be published. Required fields are marked *

GIPHY App Key not set. Please check settings

China controls the solar energy supply chain. Europe wants to change it with a huge factory in Asturias

Mercedes broke out for uncovering the anti -petitive pact