In recent days the social networks have been flooded with images showing imprisoned presidents, meetings of football players, actors, singers from different eras, or the strangest situations, but which seek to give the sensation that they are real thanks to the artificial intelligence.
The person responsible is Grok, the chatbot generated by Elon Musk’s xAI platform, and which is made to work with the social network X. Thanks to its new model called Aurora, it can generate images.
The platform presented the evolution of the chatbot called Grok2, which is now available with Aurora. The particular thing about this AI model is that it can generate images of famous people without problems.
“We trained the model with billions of examples from the Internet, which gave it deep knowledge of the world. As a result, it excels at photorealistic rendering and precisely following text instructions,” notes the xAI blog.
Its multimodal aspect also allows you to edit photos or images contributed by the user. They point out that in a week it would be available to all countries.
How does it work? First of all, to use Grok you must be a user of Free versions can also make use of Grok.
LOOK: Gemini 2.0, Google’s new thing in artificial intelligence to take advantage of its rivals
When you log in, Grok can be accessed in the left column or in the app options. The platform warns that the information provided may be false and warns that everything written will be used for AI training, including account data.
As it is a multimodal platform, Grok can be asked to generate texts, to analyze files, to tell you the news of the day.
According to the report, Grok limits its use to 25 generations (or questions to its chatbot) every two hours. If you want to do more, you must subscribe.
On the other hand, Aurora would be based on a “mixture of experts” model, which divides tasks between several specialized submodels (neural networks) to offer a combined final result.