The return of the artificial face: Google will re-enable its users to create images of people through its Gemini AI chatbot again, this after the feature was removed due to the creation of false historical images, such as Nazi figures of various races including those of African or Afro-American descent.
The upgrade will be powered by Imagen 3, the latest version of Google's AI image text generator. The technology giant quietly launched Imagen 3 through the AI Test Kitchen category, which allows testing experimental versions of new features. Now the engine will come to Gemini - the artificial intelligence engine, and will be able to create anything from a photorealistic crane to oil paintings created with the command of a few words.
Google suspended Gemini's ability to create images of people in February after users found it created false or inaccurate historical images. According to Google, the upgraded model comes with built-in protections and according to it performs well compared to other imaging models available.
The site will not allow users to create photorealistic images of public figures, content that includes minors, or violent and sexual scenes. Google plans to expand the ability to create AI-generated people to more users and languages soon. Be that as it may, the Imagen feature is different from Google's new Remagine, which allows artificial intelligence elements to be integrated into images taken in the company's new line using Google's Pixel 9 phones.
As recalled in February of this year, Meta, the owner of Facebook, Instagram and Threads, said it was promoting the development of tools that would allow it to identify images created by artificial intelligence, with the aim of combating deep fakes - the creation of false images online.
According to estimates, since 2022, almost 20 billion artificial intelligence images have been created and uploaded to the Internet, including fake photos of public or private people that were uploaded without their consent, or misinformation on a political background designed to distort the truth.
For a long time now, Matta and the other social media giants have known that they will have no choice but to do something about it. Already last year, the UK passed the Online Safety Act which makes it a crime to upload fake photos of a person without their consent.
Legislators in the US have already said in the past that they have failed in their duty to keep internet users safe, and that only legislation will oblige social networks to act to prevent the spread of fake news. Meta's move, according to estimates, will also lead other companies to build standards of trust and control over the information that is published.
The headquarters indicated that it is still not possible to identify all the content created by artificial intelligence, and that there will be those who will try to overcome the tagging technology, but said that they intend to continue looking for ways to criticize some of the uploaded content and will also call on surfers to share information about content created using artificial intelligence, so that the company can add It has a label.
In recent months, artificial intelligence images have become very sophisticated, to the point where it is sometimes difficult to recognize that they are not real. For example, last January, fake photos of the pop star Taylor Swift were uploaded to the networks, which are believed to have been created using artificial intelligence.
In Great Britain, a presentation of eight photos showing Prince William and Harry at the coronation of King Charles was distributed on Facebook, with more than 78,000 likes. One of them shows an apparent emotional hug between William and Harry after reports of a rift between the brothers. None of the eight pictures were real.
Another fake image showed former US President Donald Trump also created using artificial intelligence, after he was indicted on 13 counts of alleged election fraud.