OpenAI introduced DALL-E, a neural network that could create images from written descriptions, such as ‘an armchair in the shape of an avocado’. It was announced alongside CLIP, a model that links images and text.
What happened
- DALL-E was a 12 billion parameter version of GPT-3 trained on pairs of text and images.
- It could combine unrelated concepts, render text, and apply transformations to existing images.
- OpenAI shared examples and research findings but did not release the model publicly at first.
- Its successors, DALL-E 2 in 2022 and DALL-E 3 in 2023, brought text-to-image generation to a mass audience.
Why it mattered
DALL-E showed that generative AI could move beyond text to images, raising early questions about creativity, copyright, and synthetic media. Within two years text-to-image tools had become widely available to the public, bringing new creative uses and new risks.
Lessons for organisations
Set policies on the use of AI-generated images, including checks on intellectual property and clear labelling, and prepare staff to spot synthetic content used in fraud or disinformation. Check the terms of any image generation service for rights over the content produced.
Source: VentureBeat
Part of our Top stories archive of headline-making events in information security, privacy, and AI. If you would like help applying the lessons to your organisation, contact us.
