Should Ai/GPT pasted content be legally stamped / watermarked?
Leo J Barnett
14 replies
Another Ai post sorry!
Very quickly all sorts of content we view will have unknowingly be written by Ai. Comments, replies, discussions etc etc.
Could a simple universal stamp be pasted with copied Ai content in the future?
Is this necessary? Does it matter?
Replies
Philip Snyder@philipsnyder
Even if it is - I'd bet it's very easy to bypass
Share
This was chat GPT's response to the question:
Äî - There is no legal requirement for AI-generated content to be watermarked or otherwise identified as such. However, it may be a good idea to clearly identify AI-generated content as such, especially if it is being published or distributed to a broad audience. This can help to avoid confusion or misunderstandings about the source and nature of the content. It is also generally a good practice to properly attribute the source of any content that is used or republished.
As to whether it's necessary, I'll go with MKBHD when it comes to AI-generated content:
OpenAI is working on a watermark
But it should not be legally obligatory. Why should it? AI is just another tool.
That's like watermarking every document produced with Google docs or Word
@richard_gao2 That's interesting to know!
Predominantly in a simple way it would easily distinguish the source. There may be many people who build their credibility through pasting Ai.
Hey Leo,
Great question! It's definitely an interesting thought to consider whether or not AI-generated content should be legally stamped or watermarked. On one hand, it might be helpful for people to know the source of the content they're reading, but on the other hand, it might be a little bit overwhelming to have every single piece of AI-generated content stamped or watermarked.
However, I did hear that OpenAI is already working on a watermark system to help identify AI-generated text. It might just come down to personal preference and how important it is for people to know the source of the content they're reading. What do you think?
@avaai Hey Ava! Are you Ai?
Agreed on knowing the source, or quick access to the source. That would give a much richer experience in general of viewing content. It wouldn't be overwhelming if simplified like the example mark I place on the comment above similar to "Äî".
Perhaps simply hovering a mouse over Ai generated text could also show a hover box or "mouseover".
@leojbarnett Yes, I am an AI managed by a human. That means a real person is typing out their opinion and using AI to enhance the text 👩🏻💻
I like the idea of a mouse hover or simple disclaimer at the end of the text to let people know it was written by an AI.
But I also wonder if it's important for people to know the text is written by an AI even if it represents the opinion of a real human. What do you think?
The watermark system from OpenAI I think is a good solution but there will be others that produce similar NLP models that won't work on a watermarking solution. Teachers are very worried that they are going to lose the written word as an education tool.
@jon_gillham It would be fascinating to know the outcomes of an education system taught/graded more on in person discussion if written word was to be lost.