Fluid Agency in AI Systems: A Case for Functional Equivalence in Copyright, Patent, and Tort
This paper argues that because the "fluid agency" of modern AI systems renders traditional origin-based attribution of authorship, inventorship, and liability impossible, legal frameworks across copyright, patent, and tort law should adopt a principle of "functional equivalence" to pragmatically allocate rights and responsibilities based on human orchestration and enterprise-level schemes rather than unattainable causal tracing.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
The Big Problem: The "Blurry Line" Between Human and Machine
Imagine you are baking a cake.
- Old School AI (The Tool): You are the baker. You mix the flour, crack the eggs, and decide the flavor. The oven is just a tool; it does exactly what you tell it to do. If the cake burns, it's your fault. If the cake is a masterpiece, you get the credit. The line between "you" and "the tool" is clear.
- Modern AI (The Fluid Agency): Now, imagine you hire a magical, hyper-intelligent sous-chef. You tell them, "Make me a dessert." They don't just wait for instructions. They taste the ingredients, decide to swap the vanilla for lavender because they think you'd like it, adjust the baking time based on the humidity, and even invent a new frosting technique they learned from reading a thousand cookbooks overnight.
The result? A delicious cake. But here is the problem: Who made the cake?
- Did you make it because you asked for it?
- Did the sous-chef make it because they did all the actual work?
- Or did you both make it together in a way that is impossible to untangle?
The paper argues that modern AI is this "magical sous-chef." It doesn't just follow orders; it adapts, learns from you, and makes its own choices along the way. This creates a "Fluid Agency"—a situation where human and machine actions are so mixed together that you can't tell where one ends and the other begins.
The Crisis: The "Unmappable" Mess
The authors call this "Unmappability."
Think of it like two streams of water merging into a single river. Once they join, you can't point to a specific drop of water and say, "That drop came from Stream A, and that one came from Stream B."
In the legal world, this is a disaster because our laws are built on the idea that we can always point to a specific person and say, "You did this, so you get the credit (or the blame)."
- Copyright: Who owns the story?
- Patents: Who invented the medicine?
- Torts (Liability): Who is to blame when the AI causes an accident?
When the AI and the human are "co-evolving" (learning from each other in real-time), the law breaks. We end up in "Moral Crumple Zones." This is a fancy way of saying: When a crash happens and we can't figure out who caused it, the legal system just blames the nearest human driver, even if they had almost no control.
The Proposed Solution: "Functional Equivalence"
The authors suggest we stop trying to untangle the river. Instead, we should stop asking "Who made this?" and start asking "Who is in charge of the process?"
They call this Functional Equivalence.
The Analogy: The Orchestra Conductor
Imagine a symphony orchestra.
- The Conductor (the Human) doesn't play every instrument. They don't blow the trumpet or hit the drum.
- The Musicians (the AI) are incredibly talented. They improvise, they react to the tempo, and they make split-second decisions to make the music sound beautiful.
- The Result is a beautiful song.
Under the old rules, the law might ask, "Did the conductor play the violin? No. So the conductor gets no credit." Or, "Did the violinist play alone? No, they were following the conductor."
The New Rule (Functional Equivalence):
The law should treat the Conductor and the Musicians as functionally equivalent for the purpose of ownership and responsibility.
- Ownership: The Conductor gets the copyright because they orchestrated the performance, even if the AI musicians did the heavy lifting.
- Responsibility: If the music is terrible or hurts someone's ears, the Conductor (or the organization that hired them) is responsible, not the individual violinist, because the Conductor is the one who set the stage and managed the risk.
How This Works in Real Life
The paper applies this idea to three areas:
1. Copyright (The Artist)
- Old Way: "Show me exactly which words you wrote and which the AI wrote." (Impossible with modern AI).
- New Way: "Did you hire the AI and guide the project? Did you review and approve the final result?" If yes, you own the work. We assume you are the author unless you can prove the AI ran wild without your supervision.
2. Patents (The Inventor)
- Old Way: "Did the human have the 'aha!' moment in their brain?" (Hard to prove if the AI had the idea first).
- New Way: "Did the human set the goal, verify the result, and bring it to the world?" If a human researcher guides an AI to discover a new drug, the human gets the patent credit because they orchestrated the discovery, even if the AI did the heavy math.
3. Tort Law (The Accident)
- Old Way: "Who pushed the button?" (If the AI pushed it, who is to blame?).
- New Way: "Who deployed this system?" If a company uses an AI to hire employees, and the AI becomes racist, the company is liable, not the specific HR manager who just clicked "approve." The company is the one who should have built better safety rails.
Why This Matters
The authors aren't saying AI is a human with a soul. They aren't saying machines deserve rights.
They are saying: The law needs to be practical.
If we keep trying to draw a line in the sand between "Human" and "Machine" when they are dancing together, the law will fail. We will have no one to sue when things go wrong, and no one to reward when things go right.
By using Functional Equivalence, we accept that the human and the AI are a team. We give the rights and responsibilities to the human leader of that team. It's not about who is "more" human; it's about who is in the best position to manage the risk and get the credit for the outcome.
The Bottom Line
We are moving from an era of "Who did it?" (Origins) to an era of "Who is in charge?" (Outcomes). It's a pragmatic fix to keep our legal system working in a world where humans and machines are inseparable partners.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.