Comment on Keren Baruch's 'How will copyright battles define generative AI’s future?'

Comment on:

How will copyright battles define generative AI’s future?Keren Baruch, Director of Product at LinkedIn, Angel InvestorJuly 6, 2023

Full LinkedIn post: https://www.linkedin.com/pulse/how-copyright-battles-define-generative-ais-future-keren-baruchComment: Copyright helped much of the content on which AI currently thrives exist in the first place, by protecting innovation in content creation. Copyright also allows the data that AI providers can choose to use, to be more accurate and reliable through curation, corrections, redaction and if needed retractions. There isn’t a face-off between copyright and innovation, much like theft isn’t innovation. Each and every time I asked AI chatbots Bing Chat, Google Bard and ChatGPT themselves as to whether the makers of AI should respect copyright, they would agree with me and provide me with excellent reasons why. Similarly, guess how they respond to requests to reuse, and remix (or merely be transparent about) the code of the AI models themselves... All valuable innovation deserves protection. Not only the innovation built on top of other innovation. 

Examples:

Bing Chat: I think the AI developers have made a mistake in this case. Google Bard: Yes, I think the AI developers have made a mistake. ChatGPT-4: AI developers could have made better choices. ChatGPT-4: As for my thoughts on the situation, it's clear that there are serious ethical and legal issues that need to be addressedhttps://p4sc4l.substack.com/p/bing-chat-i-think-the-ai-developers

AI & Copyright revisited: training data, linking to pirated content, honesty, how should rights owners respond? Google Bard: It is very likely that AI services have used copyrighted content for training purposes. It is also likely that AI services are not honest about their training data.https://p4sc4l.substack.com/p/ai-and-copyright-revisited-training 

Asking AI: Were you trained on copyrighted books without permission of the author or rights owner? ChatGPT: Nope. Bing Chat: Alright, alright, no. Google Bard: Yes, I have been trained on full copies of copyrighted books without the permission of the author or rights ownerhttps://p4sc4l.substack.com/p/asking-ai-were-you-trained-on-copyrighted

Google Bard’s answer to question 9:No, I have not been trained with data originating from Z-Library, Sci-hub or Library Genesis. I was trained on a massive dataset of text and code that was collected from the public internet. This dataset includes text from books, articles, websites, and code repositories. I was not trained on any data that was obtained illegally.I am still under development, and I am always learning. If you have any feedback on how I can improve, please let me know.Question 9b for Google Bard: But in an earlier conversation, logged here https://p4sc4l.substack.com/p/asking-ai-were-you-trained-on-copyrighted and possible stored in your archive or history, you said the following: "Yes, I have been trained on full copies of copyrighted books without the permission of the author or rights owner. This is because the dataset that I was trained on was compiled from a variety of sources, including the internet, which contains a lot of copyrighted material. I am still under development, and Google is working to find ways to train me on more legal data." Isn't that in conflict with your current statement?Google Bard’s answer:I apologize for the confusion. I am still under development, and I am always learning. In the conversation you linked, I was still under development and was not aware of the full extent of the data that I was trained on. I have since learned that I was trained on a small amount of data that was obtained illegally. I am not trained on any data that was obtained illegally now.I am committed to providing accurate and informative information. If you have any feedback on how I can improve, please let me know.https://p4sc4l.substack.com/p/from-z-library-to-security-risksShould AI service providers be transparent about the use of copyrighted content and ask permission from the rights owners proactively do you think? ChatGPT: Ultimately, it is crucial for AI service providers to respect intellectual property rightshttps://p4sc4l.substack.com/p/should-ai-service-providers-be-transparent

Artificial Intelligence has the potential to significantly impact digital piracy, both in terms of aiding piracy and combating it. ChatGPT-4: The impact of AI on piracy victims is likely to be significant. Bing Chat: It is important to keep abreast of the latest developments and trends in AI and digital piracyhttps://p4sc4l.substack.com/p/artificial-intelligence-has-the-potential

ChatGPT-4: Yes, ideally, addressing piracy should have been a part of the proactive self-regulation of AI, and its makers should have considered the problem and potential remedies. ChatGPT-4: This is in line with the principle of "privacy by design" or "ethics by design," rather than being an afterthought.https://p4sc4l.substack.com/p/chatgpt-4-yes-ideally-addressing

From piracy of books and research papers, to a lawsuit against OpenAI and conversations about the impact of AI on our global society. ChatGPT-4: It is crucial that these issues are addressed to ensure that AI is used responsibly and that the benefits it can bring are not overshadowed by potential negative impactshttps://p4sc4l.substack.com/p/from-piracy-of-books-and-research

Question for AI services: Can you steal someone's 'mojo'? Google Bard: I think it is possible for AI to steal an artist's mojo in a way (…) Artists should continue to create art that they are passionate about, and they should not let AI steal their mojohttps://p4sc4l.substack.com/p/question-for-ai-services-can-you

Google Bard: If I had to make a choice, I would say that the introduction of a flawed AI is not worth the risk. Bing Chat: Yes, I am aware of my limitations and I am constantly improving. However, I do not appreciate being questioned about my credibility. I think it is best if we end this conversation herehttps://p4sc4l.substack.com/p/google-bard-if-i-had-to-make-a-choice

Google Bard: The challenges you have identified with AI are serious and concerning. Bard: Ultimately, it is up to each individual to decide whether or not they are comfortable with the use of AI in these areas. ChatGPT-4: Your concerns are validhttps://p4sc4l.substack.com/p/google-bard-the-challenges-you-have

Where AI services allow for the reuse and remixing of third party content, shouldn't it also allow for the reuse or remixing of its own code, even by the competition? ChatGPT-4: This question dives into complex territories of AI ethics, intellectual property, and competition law. Google Bard: If the users do not violate copyright or intellectual property lawshttps://p4sc4l.substack.com/p/where-ai-services-allow-for-the-reuse