Microsoft has announced the arrival of an art-generating artificial intelligence based on SLAB in following 365. This should greatly facilitate the creation of illustrations but this announcement raises several questions as this technology is controversial.
This news is not so surprising. In 2019, Microsoft invested a billion dollars in OpenAI, the firm behind DALL-E. Also, the queue to access the AI was removed a few weeks ago.
A new application based on DALL-E and called Microsoft Designer
The Redmond firm took advantage of the Microsoft Surface Event 2022 to announce the arrival of the Microsoft Designer application. This will work in the same way as other art-generating AIs: the user will only have to type in a few keywords to instantly create ” a variety of designs “.
The tool will be available as a free standalone app and a more feature-rich version available to paid Microsoft 365 subscribers. or illustrations for PowerPoint presentations. However, for now, Microsoft Designer is not optimized for full deployment and is only available in a limited web version.
Additionally, Microsoft plans to bring DALL-E support to its Bing search engine as well as its Edge browser, “ so that you can use your words not only to search but also to create “. This feature will be called Image Creator.
This technology is controversial
If art-generating AIs are on the rise, their large-scale deployment as computer products raises questions on several points. Indeed, technologies like DALL-E have been trained on images from the Internet; among them, works created by artists, photographers and graphic designers who believe that their work is being stolen. Moreover, online communities have decided to attack art-generating AI directly, and the Getty Images platform has simply decided to ban the distribution and purchase of this type of content.
These AIs can also be used for malicious purposes. The latest example is that of Stable Diffusion, an art-generating AI available in open-source, which has been exploited to create pornographic content. On this point, Microsoft wants to be reassuring and ensures that OpenAI has filtered “ the explicit sexual and violent content of the data set used to train the model » and that she also « deployed filters to limit the generation of images that violate the content policy “. In addition, the company has set up additional blocking of queries on sensitive topics “.
Are these filters fully waterproof? Only the future will tell us.
Sources: The Verge, Engadget
9