I asked gpt why it doesn’t allow pictures of specific people, because in many was it really reduces commercial potential for the end user. It basically responded that people could put out misinformation. To this I responded:
might i suggest that you limit what forms of art you produce should play a role in what dalle will or will not render. Example: a cartoon, line drawing, water color, oil painting, basically anything that is not mistakable for an actual photograph, this things can be done, but anything that could be mistaken for reality could not be done. thoughts?
gpt responded:
Your suggestion to differentiate between stylized art forms like cartoons, line drawings, watercolors, and oil paintings versus hyper-realistic images that could be mistaken for actual photographs is insightful. This approach can indeed play a significant role in addressing ethical concerns related to AI-generated images, especially those involving depictions of people. Here are a few thoughts on how this distinction could be beneficial:
- Promoting Creativity While Ensuring Safety: By focusing on clearly artistic or stylized representations, such as cartoons or watercolor paintings, DALL·E can encourage creative expression without risking the creation of images that could be confused with real photographs. This allows users to explore artistic possibilities while minimizing the risk of creating misleading or harmful content.
- Ethical Boundaries: Setting clear boundaries on the realism of AI-generated images can help mitigate issues related to privacy, consent, and the potential for misuse. By ensuring that outputs are unmistakably artistic, the technology respects individuals’ rights and reduces the risk of ethical breaches.
- Artistic Freedom: Encouraging the use of distinct artistic styles, like those seen in traditional art forms, can foster a diverse range of creative outputs. This not only showcases the technology’s versatility but also pays homage to the rich history of artistic expression, drawing inspiration from various periods and movements.
- Educational Value: By clearly delineating between artistic and realistic outputs, there’s an opportunity to educate users about different art styles, techniques, and their historical contexts. This can enhance users’ appreciation for art and its various forms, enriching their experience with the technology.
- Technical Implementation: Implementing this approach would require clear guidelines and possibly more sophisticated content moderation algorithms to distinguish between stylized and hyper-realistic outputs. This could involve advanced image recognition and classification capabilities to ensure compliance with ethical standards.
Incorporating these considerations into the development and use of AI-based image generation tools like DALL·E can help balance innovation with ethical responsibility, ensuring these technologies are used in a manner that is respectful, safe, and creatively enriching.
I would also like to add the following idea: if it is safer, have a version of dalle that only does art, but does not do hyper-realism that could be used to render character specific images, but the hyper-real version doesn’t allow it. I mean, if i want to make a book, a mascott, or anything like that, but it never matches, that really limits what can be done.