Watch On:
Summary
AI can now create any image in seconds, bringing wonder and danger DALL-E. Named for Salvador Dali and Pixar’s WALL-E, DALL-E creates images based on prompts such as: “A hobbit house designed by Zaha Hadid.”
The introduction of DALL-E has triggered an explosion of text-to-image generators. Rival start-ups soon went public, including Stable Diffusion and Midjourney, which created the image that sparked controversy in August when it won an art competition at the Colorado State Fair. Researchers worry that these systems produce images that can cause a range of harms, such as reinforcing racial and gender stereotypes or plagiarizing artists whose work was siphoned without their consent. Fake photos could be used to enable bullying and harassment or create disinformation that looks real. said UC Berkeley professor Hany Farid, who specializes in digital forensics, computer vision, and misinformation. “More recently, advances in AI gave rise to deepfakes, a broad term that covers any AI-synthesized media from doctored videos where one person’s head has been placed on another person’s body to surprisingly lifelike “photographs” of people who don’t exist. OpenAI wanted its AI to benefit the world and act as a safeguard against superhuman AI in the hands of a monopolistic corporation or foreign government. At the same time, OpenAI tried to combine vision and language, training GPT-3 to find patterns and links between words and images by ingesting massive data sets scraped from the internet that contain millions of images paired with text captions. Whether the system sees an avocado, hears the word “avocado,” or reads the word “avocado,” the concept that gets triggered should be exactly the same.
Show Notes
AI can now create any image in seconds, bringing wonder and danger. The introduction of DALL-E has triggered an explosion of text-to-image generators. Each evolution of image technology has introduced potential harms alongside increased efficiency. Since DALL-E’s outputs are in images, OpenAI can view how the system represents concepts. Most of the team’s observations revolved around images DALL-E generated of photorealistic people, since they had an obvious social impact.
Source
https://www.washingtonpost.com/technology/interactive/2022/artificial-intelligence-images-dall-e/
