Google is carefully restarting AI image maker Gemini

This article was last updated on August 29, 2024

Canada: Oye! Times readers Get FREE $30 to spend on Amazon, Walmart…
USA: Oye! Times readers Get FREE $30 to spend on Amazon, Walmart…

Google is carefully restarting AI image maker Gemini

Within a few days, AI image generator Google Gemini can generate artificial images of people again. According to Google, this is an improved version that is currently only available to paying users.

Gemini can, among other things, generate artificial images based on textual descriptions from users. This makes it similar to previously released technologies from competitors, such as image generator DALL-E and Bing Image Creator from Microsoft.

Taken offline

Six months ago Google took the ability to generate images of (non-existent) people offline, after criticism from users. The tech giant wanted to counterbalance stereotypes according to research sneak into artificial intelligence technology on a regular basis. For example, on other services, when a user asked for an image of a ‘prisoner’, the majority of images of black people were shown.

Google Gemini wanted to do better, but failed: it turned out to be virtually impossible for users to have images of white people made. For example, the founding fathers of the United States, Nazi soldiers and Vikings were historically inaccurately depicted as black or Asian people.

It led to hilarity on social media, where users tried everything to create images with white people in them.

Taking Gemini partly offline was a painful decision for Google. The original intention was to suspend the service for ‘a few weeks’. It now appears that it took more than six months to implement improvements.

Elon Musk

Google is committed to being at the forefront of developments in artificial intelligence. Yet in practice, it has been lagging behind competitors’ AI innovations for some time. For example, the tech giant was surprised by the rapid rise of ChatGPT, from OpenAI.

While Gemini was partially unavailable, Elon Musk’s company xAI announced one such 6 billion dollars in investments. This allowed it to launch its own AI model Grok, which can also be used to generate images. There was different criticism than Gemini: too many things were allowed with Grok.

For example, Musk’s technology appears to produce violent and copyrighted images can be made. Pictures of Mickey Mouse or famous artists with weapons and drugs have been circulating on social media for weeks.

‘Not perfect’

Google is playing a trick with the ‘reopening’ of Gemini’s image generator. While the technology can’t imitate real people or generate violent or sexually explicit images, Gemini “will not be perfect, just like any other generative AI tool.”

It is not yet clear when Gemini’s image generator will also become available to non-paying users.