Open AI has decided to take decisive action against the unfair competition they have created. Dall-E 3, from now on, will add a watermark that will allow clearly distinguishing which images were created by real artists, and who were generated by those who believed they were artists by adding commands to an AI.
In this way, Open AI tries to solve one of the most pressing problems today regarding generative Artificial Intelligence images: that users who generate images with these tools cannot replace the work involved in a long experience and many hours of work on an artistic piece.
Dall-E 3 against misuse
As Open AI has publicly explained, they have decided to add two elements to the images generated by AI. One of them would be the watermark on the image, a resource that will make the images clearly identifiable as AI-generated. The other one, the C2PA format metadata, which will provide information about this image generation.
These systems, really, will not be difficult to avoid for those who want to continue carrying out their generative images, but it will also mean a loss of quality in the image and, consequently, a loss of value as generated content.
The dilemma of generative AI
Generative AI has had many problems related to this issue from the beginning. Especially, since the jump to GPT-3 iteration, which represented a big qualitative leap, systems like ChatGPT or Dall-E started to be massively used by users, either in a leisurely way or trying to emulate the work of other people, even claiming that it takes “a lot of effort” to choose the right words for these Artificial Intelligences to do a better job. However, their effort is still infinitely inferior to that of a true professional.
Therefore, thanks to changes like this measure carried out by Open AI, those who particularly use their tools to generate images and pass them off as truly original artistic creations will see their system truncated. In fact, the European Union is also making efforts to legislate this matter properly so that citizens are adequately protected against certain uses and abuses by AI.
