When Artificial Intelligence Clashes with Intellectual Property

Once again, the world finds itself at a crossroads

Historical Underpinnings (Humans and Machines a Tricky Alliance)

Several ideas are at play here. The first is a human vs. machine when we moved from subsistence farming to commercial. We have been here before in the sense that when we transitioned from very basic agriculture where we grew only what we needed to eat and planned for seasons based on these changes everything was quite basic. When however, we began to gather in communities and expand, our needs changed. We bought horses and introduced commerce and the exchange of money and currency. But something else also changed. We started depending less on humans and more on machines. Horses could do a lot more than men. For other larger measures, we expanded and began to trade in our fellow humans. That was between the industrialization and the major shifts that took place in agriculture. 

The changes that were faced when industrialists brought in factories are the same changes we are facing as we fear that we are gradually being replaced. 

And that is what AI, Robotics and the closing of the gap between human and machine seems to present. 

The writer, the machinist, the cook, the waiter, the team that sweeps the road, the painter, the shoemaker, the lawyer...heck even the president could one day be replaced by a system that uses what we have fed it through our posts and what it has learnt about human in the last few centuries to govern us. 

You could call this "Aicracy". 

But this article is really about the recent AI and IP case involving one or more LLMs which are being accused of illegally training their models on the information that has been available online from authors of all sorts of shapes and colors. 

What is at stake though? What does AI have to do with IP and Copyright and what do the scholars or books say about this issue.

The Nitty Gritty (HIstory, Ideas Vs. Expressions, Geographies and Theories, The Non Human Claim, Fair use doctrine and the decline of formalities) 

The First has to do with definitions. In a CopyrightX course i took in 2015 through of Harvard Law School and the Berkman Klein Center for Internet and Society, CEHURD and MUK (MUCIT) and taught by William W. Fisher the WilmerHale Professor of Intellectual Property Law Faculty Co-Director, Berkman Klein Center for Internet and Society. 

For Access to some of the materials and lessons use this youtube link 

We learnt that you cannot protect ideas what you can protect are expressions.

For example, everyone can write about the father and a son story with some betrayal but what makes it unique are key aspects in the book or the song or how this story is told that can be captured to show this distinctiveness. Have you noticed for example how many different movies have explored the idea of a man or villain going through a security check?

When and What-history and copyright law

The second is at least in the study was an understanding of the historical context of these laws and the rapid or sometimes not so rapid development over the years. Software only emerges in the late seventies or thereabouts as something that can be protected. We see major complexities here because of the 'tonnes' of code that are found in even the most basic software systems. 

In the first section we demonstrated that for something to be unique and therefore copy protected you must have a means of capturing that uniqueness. This is impossible with closed systems. In others however there has been a proliferation of a movement that has favored the creation of material Open Source Software (Use this link to access a conversation we shared with Data Umbrella) where the belief is that this stuff should be free and open and that from the Open Source code developers are allowed to build and make adjustments. 

What's Geog gotta to do with it

The third ideas is a merger of two thoughts the first is the geographical concerns which also falls into the four major theories of copyright. What type of law does your country use? Common law based on judicial decision verses Civil law (or statutory law) which is based on Statutes or written laws passed by legislative What theory based on that law is applicable in views that your country will adopt with regards to IP law? A quick question to ask yourself is why does it seem that EU or European Law seem to always at least in general rule against American Companies (in a non-political way)?

The Fourth (Click here to prove you are not a robot) 

Then there is the rather insidious claim of the non-human nature of AI. Who do you punish for the evil undertaken by AI in copying someone's material? Here is a story. Have you ever visited certain Channels on Telegram and wondered how so many movies are available for you at no cost? While it is possible to ban a human from making use of the videos, movies and series, it is harder to make the case for a bot. Almost all laws take humans into context but how do you charge a robot for IP violations. Are bots obligated to read the agreements.  

Fifth (In and out of Public domain)

When you study the progress of IP in the US you will notice that there is a timeline upon which all these protections run and when this expires you and are not renewed, these fall into the public domain. AIs can simply be triggered to grab as many IP products as possible when they expire and therefore take them into ownership. Since is not human can it claim access to these products? They may not even need to go that far. Just scour the internet for products just as they land in public domain and then use them as LLMs. Here is a list i found online 

Winnie the Pooh and Bambi both saw their copyrights expire last year, and they’ll soon be followed by the likes of Peter Pan (2023), Mickey Mouse/Steamboat Willie (2024), Popeye (2025), Superman (2034), Batman (2035), and Buggs Bunny (2035). The link to follow is here. 

Sixth (Content and use of that content)

How do LLMs work is it fair to just copy structure only, is it fair to just use beginning and end. Is fair to use data analytics to find key parts that make a book successful or other tools and then mine that data for insights? The definition for LLMs is found in Geek For Geeks.