Google has repeatedly shown what its various machine learning algorithms are capable of, such as MUM either LaMDAand continues to mirror them with a new model of artificial intelligence called ‘Image’.
According to Jeff Dean, the company’s head of AI, this promises to “unleash joint creativity between humans and computers” Y be able to generate images from a single, short and simple text description.
‘Image’ is very similar to DALL-E-2, artificial intelligence developed by Open AI (company founded by Elon Musk) which also allows you to create an image from a written description. However, there are some differences between the two models, such as the level of detail and the efficiency in creating this image.
Specific, Google claims that its AI provides results with a much more precise level of detail than other systems. To verify this, the company has created a benchmark called drawbench, which compares your AI model with similar models, and presents the results side by side for the user to view. The “human evaluator” can distinguish between them and choose the most realistic.
These people, according to the company, concluded that the photographs produced by ‘Imagen’ were of higher quality and “images aligned to the described text”, being better than the rest of the models.
An artificial intelligence more efficient than others
‘Image’, Google’s AI that generates an image from a short text description, is also “more computationally efficient, more memory efficient, and converges faster” thanks to its proprietary architecture called AT. As a result, surreal images are produced more accurately than any other model and any type of text description.
“A transparent sculpture of a duck made of glass. The sculpture is in front of a painting of a landscape”, “a couple of robots having dinner with the Eiffel Tower in the background”, or “a dog looking curiously in the mirror, seeing a cat”, These are phrases that Google uses as an example to demonstrate what its AI model is capable of.
These are some of the ones that can be found in your Web page:
Google, on the other hand, claims that “Image” can also generate images with descriptions based on specific locations or even complex text. For example, if a user enters “A giraffe running with a rabbit in the La Bombonera stadium in Buenos Aires”, the company’s AI will generate an image based on this description and will understand the scientific name of both animals, as well as the places.
‘Image’ is, at this time, an internal project and not accessible to the public, as it may lead to the creation of images containing “stereotypes and harmful representations”, as explained by the company.
“Image relies on text encoders trained on uncurated web-scale data and thus inherits the social biases and limitations of large linguistic models. As such, there is a risk that Imagen has encoded harmful stereotypes and representations, which guides our decision not to release Imagen for public use without further safeguards.” mention Google.
How to see what Google account data apps access
From the mobile:
1. Open Gmail.
2. Press on the three horizontal stripes of the top margin.
3. Go to Setting.
4. Click on the account to be configured (if there is more than one added on the mobile).
5. Click on Manage your Google Account.
6. Swipe right to the option Security.
7. Finally, enter the section Third party apps that can access your account.
– From a Web navigator:
1. On a computer, press on the dotted square that appears in the upper right margin next to the user’s name.
2. Enter Bill.
3. On the right margin press where it says Security.
4. Finally, press on Third Party Apps that can access your account.