2mo ago

Are we at risk of losing classical editions of texts to AI?

How will we know that we are reading something authentic written by the original author vs something that has been altered and rewritten? This may have been a problem in the pre-AI world as well, but with the ease one can generate text now I wonder if there is a much greater risk.

14 comments

No. If somehow someone starts to try that is as easy as digitally signing the original documents.
- It’s a pity that blockchain technology is so intertwined with crypto—it would make a great anti-AI check. As in, a way for a decentralized community to maintain an expanding collection of information while efficiently verifying that no existing works have been altered.
  Sure, you can sign individual documents—but then for each document you need a separate channel to verify the original signature.
  
  Friend, PGP signed messages were around in the 90s. Key signing parties. Web of trust.
What is authentically and originally written is impossible for ai to replace. Write with a quill and ink via candlelight. What is possible is editing. Take your comfort where you find it there. Editing has been changing original manuscripts for centuries for better and for worse
Same way as always? Consider the source.
No. We are not.
I can imagine the EU mandating AI in the browser and social apps. To keep the kids safe, ofcourse.
It will scan all the content, and modify it to a safe version, for your benefit.
"Voting started for ~~Catalan referendum on independance~~ The X factor Spain!"
- Ah yes, the EU, foremost actor in stealthy propaganda. 🙄
  
  I'm describing a next iteration of chatcontrol. The EU already proposes to have LLMs in every social app to see if you're not communicating illegal thoughts.
  Only a small change to have it alter content.
Not yet. But soon (10y maybe in my guess)

14 comments