A game to understand artificial intelligence DALL-E

1662804551 138 mano dibujando.jpg
1662804551 138 mano dibujando.jpg
Share on Facebook Share on LinkedIn Share on Pinterest Share on Email Share on Telegram Share on WhatsApp

In recent months, artificial intelligences (AI) have proliferated, such as DALL-E, capable of generating images from simple text instructions (prompts) that indicate which subjects or objects should appear in the resulting image.

Finding out how an AI converts instructions into images can help to achieve a better generation of these types of images.

The results are usually surprising and depending on the description they even result in amazing “creations”. The basis of the operation of this type of AI is the handling of large amounts of images of which there is a description of what appears in them. Upon receiving the AI the description in natural text form it contrasts the entered words with the images that correspond to the ones it has in its database, it contrasts it and creates a new image from the recognized elements.

When it comes to obtaining the best results, it is not only necessary to write some prompts but also to know in detail the operation of the process by which AI converts words into images. Or, strictly speaking, knowing how the previous process works, how from previous images for which a description is available, this allows the AI ​​to learn to relate one to the other when it is asked to translate words into images.

DALL-Edle game

As a game but also as a way to learn about this process, it has appeared DALL-Edle, a web page that shows different images adding four text options among which you have to select the one that is considered to have served as prompt to create these images.

Five sets of images are proposed with four possible texts that would correspond to the prompts that would have generated them, of which one will have to be selected. After marking the corresponding option, DALL-Edle reports that how many hits has been taken in this peculiar test of recognition.

DALL-Edle catches

DALL-E is one of the most popular AI imaging models and is the one based on DALL-Edle to try to make users understand how their model works. To show the playful part, they have taken as part of the structure of the name of the web not only that of DALL-E but also that of the popular game Wordle in which you have to find a word in a maximum of 6 attempts.

The mechanics of this game proposed by DALL-Edle tries to make the player capable of understand how the process by which the AI ​​transforms instructions or descriptions into images works. In a way, it is about confronting human creativity with artificial “creativity” and demonstrating whether man is capable of understanding the machine and vice versa.

Previous articleThis is what will happen to you if you don’t use your Paypal account for a long time
Next articleThe “Sims 5” is coming to Android and iOS phones
Brian Adam
Professional Blogger, V logger, traveler and explorer of new horizons.