OpenAI’s DALL-E 2 AI is Bad News for Artists
OpenAI’s DALL-E 2 shocks those who believed that artificial intelligence would not enter the realm of creativity. Is DALL-E 2 here for artists’ jobs?
How does DALL-E 2 work?
DALL-E 2’s impressive visuals almost seem like magic. However, the details of how it creates these stunning and realistic images are easy to grasp.
DALL-E 2 has two components. GPT-3 is the first and the most advanced natural-language machine learning algorithm currently available. DALL-E 2 uses the CLIP (Contrastive Language Image Pre-training) OpenAI model.
GPT-3, CLIP, and CLIP enable a computer to understand complex natural language and create it. It learns relationships between concepts by training the DALLE neural net with billions upon billions of images and their natural language descriptions (mainly) from the internet.
DALL-E, in a sense, is the opposite of standard machine learning practices, where you provide an image, and the AI attempts description.
You can think of the famous “Not a Hotdog” app from the TV series Silicon Valley. This is because instead of asking the AI if the photo is a hotdog, you describe the hotdog to it, and it generates an…
