Ethical Differences Between AI-Generated Text and Art: Attribution, Transparency, and Creative Intent

Type B:

Talk briefly about what ethical differences, if any, exist between using AI models trained on text to create text vs using AI models trained on art to create art

The ethical differences between using AI models trained on text to create text and using AI models trained on art to create art are primarily related to issues of attribution, transparency, and the preservation of creative intent:

  • Attribution and Ownership:
    • Text AI: When using AI models to generate text, it is crucial to attribute the sources and data used for training. Ethical concerns may arise if the AI generates text that closely resembles existing works without proper citation or if it infringes on copyright.
    • Art AI: Similar to text AI, the use of AI models to create art should also respect copyright and the sources of artistic inspiration. Proper attribution and acknowledgment of influences are important to maintain ethical practices.
  • Transparency:
    • Text AI: The use of AI for generating text should be transparent, and the audience should be aware that the content is machine-generated. Transparency helps maintain trust and ethical standards.
    • Art AI: Transparency is equally important in the context of AI-generated art. Viewers should know that AI played a role in the creation of the artwork. Hiding the fact that a piece of art was generated by AI can be seen as deceptive.
  • Creative Intent:
    • Text AI: While text AI can assist with content creation, it should be used as a tool that complements human creativity and not as a means to manipulate or misrepresent information. The ethical use of text AI involves preserving the writer’s intent and voice.
    • Art AI: The ethical use of AI in art involves ensuring that the human artist’s creative intent is respected. AI can be a collaborator, but it should not overshadow the artist’s unique vision and expression.

In summary, the ethical considerations when using AI models trained on text or art revolve around issues of transparency, attribution, and the relationship between human creativity and machine assistance. Both text and art generated by AI should be used in ways that respect the rights of original creators and maintain a clear line between human and machine contributions.

Title generated by ChatGPT

My thoughts:

I think I still don’t understand how citation might work in regards to AI. ChatGPT seems to think you need “proper attribution and acknowledgement of influences” when using either AI text or art, but that’s not something it gives you for each individual output. I don’t know if there’s a way to know all the text or images an AI is trained on, but unless there is AI doesn’t credit its sources at all.

For art in particular, it’s one thing to be inspired by an artist’s style, and acknowledging that influence is a courtesy rather than a requirement. However, it’s another thing entirely to actually use their own work to create a new artwork in their style, like some AIs do, without acknowledgement or permission at all. As was said by Kirby Ferguson in Everything is a Remix, “AI is derivative by design and inventive by chance”. It can create, but it’s not creative in the way that actual human artists are and it can’t draw influence and inspiration from other artists the way actual human artists can. It essentially contorts actual work, which is something that is, in other circumstances, generally protected by copyright. (Popular artists can’t sample a few seconds of a song uncredited without a scandal blowing up around them.) Copyright is meant to give artists control over what happens to their art, but once it’s absorbed into an AI model, they lose that control. I get why artists would be really upset by that.

Image by William F. Hertha,

4 thoughts on “Ethical Differences Between AI-Generated Text and Art: Attribution, Transparency, and Creative Intent

  1. Catherine, I generally agree with you. But I wonder about your assertion that ” it’s another thing entirely to actually use their own work to create a new artwork in their style.” What about collage or mixed media art? When I’ve seen stuff like that in the MOA or other galleries, the artists do not give citations for where they found their materials.

  2. Yeah, I also thought it was interesting that the LLM stated that it is unethical to use sources without proper citation. I have yet to see real citation let alone proper citation.

  3. I found the attribution and ownership section that your AI wrote very hypocritical, as I have yet to see ChatGPT credit anyone. (Though I don’t know if an AI can be hypocritical if they don’t have the power of choice.) Additionally, often there isn’t even a way for someone using AI to find out where the AI is gathering their material.

  4. I think its interesting that AI makes sure to say that the user needs to understand that it is AI generated and to check the facts when it is questioned about citations. It makes me wonder where the liability falls when citing AI’s work. I also found it ironic AI was saying that it’s art is supposed to be collaborative, when artists are worried about it replacing them.

Leave a Reply

Your email address will not be published. Required fields are marked *

Proudly powered by WordPress | Theme: Lean Blog by Crimson Themes.