This is a bipartisan effort. This is about AI. This is about the economics involved. This is forging a consensus that is actually easy to reach.
2 Likes
I think this is fundamentally a copyright issue. Could AI take something like Huckleberry Finn and redo it just beyond copyright rules? Easy for AI to do. What are the rules for derivative art?
AI can absorb ideas/ information from everywhere and weave it into something new but containing the same ideas. What are the limits?
2 Likes
The limits are those of the AI “imagination” making.
Can a human read 10 or 100 popular fictional crime books and then after having done that and learned the various techniques used for fictional crime writing, to write their own fictional crime book using that learned knowledge?
And if a human can do it (they can and do regularly), can the human use a tool (either a simple one or one that runs on a big fancy computer) to assist with the task? For example, let’s say the tool used is Notepad or Word, so while reading those books, the human takes notes on their computer. Notes like:
- First name of detective should always be 4 or 5 letters with only two syllables.
- Last name should always be longer than first name.
- Name should be unique in some way to make it memorable.
- Always have a sidekick of some sort.
- Story must have at least 5 major steps between the crime and the solving of the crime. Each step should have at last 3 turning points.
- Etc.
Later on, while writing their own book, the human uses all these notes to ascertain that they are writing properly and with sufficient content to make it interesting to a typical reader.
1 Like
I’ve read about AI detection and the challenges it presents in distinguishing between AI-generated content and human-created works. I think the concern about AI redoing classics just beyond copyright rules might be valid. AI can absorb vast amounts of data and generate new content that may closely resemble existing works. However, the limits of what AI can do, and the legal implications, seem to revolve around the concept of “derivative work.” If a human can take inspiration from various sources to create something new, the question is whether AI, as a tool, can do the same. The key issue is how we define originality and where we draw the line between inspiration and replication.
1 Like
The courts are considering ruling in terms of degrees or percentages of infringement and requiring royalties for a newspaper like the NYT that is a rip-off. The royalties will be tiny. It breaks their copyright without their permission.
This is not a new problem. Architects tell you if someone takes their design and changes one line they can claim its new and subject to its own copyright. There must be rules on exactly how similar it must be.
Take a volume, run it through AI, capture the plot and outline. Then rewrite it with new language or slang. Is that subject to copyright or not?
Music industry. Piano has 88 keys. How many possible combinations are there and which ones are copyrighted?
AI makes these variations easy and probably numerous but the rules involved have been around for decades at least.
1 Like
If you just play with one finger and play 10 notes with a constant beat you have 27,850,097,600,940,212,224 choices. But I suspect most of them don’t have a pleasing sound. If the notes were architectural structures most would fall down. 
Mike
The problem with deciding degrees is the solution. The courts are considering that if there are 10 sources for an article by AI depending on how much of each source is used that will be the payout in royalties. This can mean word for word which would allow AI to reword things to avoid payouts of royalties.
Meanwhile the NYT or WaPo will have their news stories that are expensive to produce reproduced across the web at no cost in real terms.
The only issue is opting out. If this is like the DMCA there will be no opt out. The royalties would be destructive of the actual author’s copyrights. Tiny royalties would undermine the ability to produce the product.
2 Likes
Sometimes I truly wonder what my industry is doing to society. 
You do not want to hear this. It’s not just printers. This is endemic of the tech industry.
Colin McEnroe the presenter is a very colorful figure in CT.
Re: news stories reproduced across the web
You can copyright the photo of a tree. You cannot copyright the tree. Someone else can take a photo of the same tree and have their own copyright.
It can be argued that a news story is copyrighted but not the facts in it.
1 Like
It can be argued that the laziest bum will click a button and add AI news to his blog most of which word for word is not his or even AI’s.
We have Google doing that now at the top of most results.
The problem is when no one pays reporters do we have anything left in civic life? Or business life? Hear the story on printers and you might do a 180.
Top music acts need to do stage appearances now to make their main money. Going into the studio is a waste of time. Spotify and others do not pay properly.
1 Like
Oh no! You forced it! You made me post this. It doesn’t take 10 notes, only four cords.
(caution, coarse Aussie language)
3 Likes