AI-generated text should be detectable, but Apple needs to avoid Anthropic’s huge error
Ben Lovejoy argues that AI-generated text should be detectable, but Apple needs to avoid Anthropic's huge error of deliberately mangling language to mark it as AI content. He hopes Apple will choose a smarter path than that employed by Anthropic.
Intelligence analysis by Llama
Ben Lovejoy thinks AI-generated text should be detectable, but Apple should avoid Anthropic's approach of deliberately changing human-written text to mark it as AI content. He hopes Apple will choose a smarter path.
Imagine you write a story, and then a computer helps you fix any mistakes. But what if the computer changes some of the words to make it look like it wrote the whole thing? That's not fair, and it's not what we want. We want to be able to tell if a story was written by a human or a computer.
Analysis
The EU Initiative on AI-Generated Text Watermarking
The European Union has recently introduced an initiative to digitally watermark AI-generated text, similar to the approach used for AI-generated images. This move aims to combat the proliferation of AI-generated content, which can be difficult to distinguish from human-written text. The initiative requires AI companies to embed invisible characters or language patterns in their text output to allow for detection, even when the text is copied and pasted.
Anthropic's Perverse Approach
Anthropic, a prominent AI company, has announced that it will comply with the EU initiative by embedding language patterns in its AI-generated text. However, the company's approach has been met with criticism, as it involves deliberately mangling human-written text to mark it as AI content. This raises questions about the ethics of AI-generated content and the need for a more nuanced approach.
The Problem with Anthropic's Approach
The issue with Anthropic's approach is that it involves changing the wording chosen by human authors who have asked AI systems for proofreading. This is not only unethical but also unnecessary, as a 'Proofread by AI' label would be sufficient. The company's approach is also problematic because it can lead to the creation of AI-generated content that is indistinguishable from human-written text.
Apple's Response
Apple will need to respond to this issue in the near future. The company has already prepared a feature called Apple Reference Image to combat AI-generated imagery. It is likely that Apple will need to do something similar with Siri AI tools, which can be used for proofreading and writing. The company should choose a smarter path than Anthropic's approach, which involves deliberately mangling human-written text.
Conclusion
The growing use of AI chatbots to assist with writing and editing raises tricky questions about what constitutes AI-generated content. Ben Lovejoy's opinion piece highlights the importance of detecting AI-generated text and the need for a smarter approach. Apple should choose a path that prioritizes the integrity of human-written content and avoids the pitfalls of Anthropic's approach.
Key points
- AI-generated text should be detectable and marked as such
- Anthropic's approach of deliberately mangling human-written text is problematic
- Apple should choose a smarter path than Anthropic's approach
- The growing use of AI chatbots raises tricky questions about what constitutes AI-generated content
If Apple chooses a smarter path than Anthropic's approach, it could lead to a more transparent and trustworthy AI-generated content ecosystem. This could also encourage more human authors to use AI tools for proofreading and writing, leading to better quality content.
If Apple follows Anthropic's approach, it could lead to a proliferation of AI-generated content that is indistinguishable from human-written text. This could undermine the integrity of human-written content and lead to a loss of trust in the AI-generated content ecosystem.
