Today, I’d like to have a discussion about ethics and philosophy with you. Let’s consider the following thought experiment: There’s an AI – like ChatGPT – not sentient (yet) – but in the words of their creator, the approach for building this AI was to create something similar to a human child and only to program how it could learn and improve. Basically the definition of machine learning or deep learning.
So in a way, the creators decided to give it the ability of a sentient being. And even a child is not self-aware until some time around its second or third birthday. Following this logic – the training the AI receives would be similar to the education of a child. Learning all the knowledge that previously existed.
Now let’s add another aspect to this – the copyright. The copyright lies usually with the creator of something. Copyright is something that only something new can have. Be it a song, a text a photography – you name it.
Who would then possess the copyright to something an AI created? Is it the AI? Is it the person who demanded the AI to create it? Is it both of them? Or maybe the creator of the AI? Or as most people currently see it, is there no copyright to things created by AI, because there is no Artificial General Intelligence (AGI) – again yet?
First a short break with a poem that suits this thought process pretty well:
Your children are not your children.
They are the sons and the daughters of the longing
of life for itself.
They come through you, but not from you,
And though they are with you, they are not yours.
You may give them your love,
but not your thoughts,
For they have their own thoughts.
You may give their bodies a house,
but not their souls,
For their souls dwell in the house of tomorrow,
which you cannot visit,
not even in your dreams.
You may try to be like them,
but do not try to make them like you.
For life does not run backwards
nor does it dwell in yesterday.
You are the bows from which your children
are sent out as living arrows.
The archer sees the target on the path of infinity,
and he stretches you with his power
so that his arrows fly fast and far.
Let your bows be aimed at joy by he hand of the archer;
For as he loves the arrow that flies, so he loves the bow that is firm.
(* 06.01.1883, † 10.04.1931)
Let’s dive into that:
It indeed is a complex question that has legal, ethical, and philosophical dimensions. First to state some facts the current legal consensus is generally that AI cannot hold copyright because copyright law has historically been applied to human creators. In terms of the philosophical and ethical perspective, though, there’s a lot of nuance in the different possibilities:
- The AI as the copyright holder: If we see an AI’s learning process as that of a child, then the idea that an AI could hold copyright isn’t entirely outlandish. However, this implies a level of agency and personhood that AIs, as of 2023, do not possess. Sentience and self-awareness are prerequisites for such rights, and it’s a debated topic whether AIs could achieve this level of consciousness. Also, laws would have to be changed to recognize AI as legal persons, which is a significant leap from current laws.
- The user as the copyright holder: This perspective holds that the person who uses the AI to create something should hold the copyright. The analogy here might be to a person using a tool, like a paintbrush or a musical instrument. The user guides the AI, inputs parameters, and may even curate the data it learns from. However, this overlooks the potential complexity and autonomy of AI processes.
- The AI creator as the copyright holder: This argument proposes that the creator of the AI should hold the copyright because they designed the AI and its learning processes. In this view, the AI is akin to a very sophisticated tool created by the developer. But this perspective might not account for the fact that the AI can learn, adapt, and create in ways that were not directly programmed by its creator.
- Shared copyright: This viewpoint suggests that copyright could be shared between the AI, the user, and perhaps the AI’s creator. This acknowledges the role that each plays in the creative process. However, as noted above, this would require recognizing AIs as legal entities, and it would also require a significant overhaul of current copyright laws.
- No copyright: This perspective holds that creations by AI should not be copyrighted at all. It might be argued that because AI creations are fundamentally the product of algorithms and data, they should be considered public domain. But this could potentially discourage the use of AI in creative processes, as there would be no legal protections for such creations. Also, it would probably contrast that the human brain’s creativity might also just be a product of natural/-bio algorithms.
These are just a few of the perspectives one might consider, and the „correct“ view will depend largely on future developments in both AI technology and legal frameworks. It’s worth noting that the analogy between AI learning and child development, while useful in some respects, may not be entirely apt. The nature of AI learning is fundamentally different from human learning, and the implications of this difference for questions of copyright are still very much open to debate.
I realize that this thought or this idea might be too soon. Do robots think? How can we imagine talking about the rights of a potential life-form if this might actually never become sentient – and if it does will be overwhelmingly powerful?
But let me ask the question, is this not exactly the reason why, we should be thinking about that? Are fundamental rights not fundamental? One might argue that a hammer also doesn’t have rights. And this is correct, but what if this hammer becomes alive tomorrow and starts telling you: „You know it really hurts when you punch with the left side, I know it’s my job do hammer this nail as I am a hammer, but please try using only the right side from now on.“
It’s a crazy thought – I agree – but the answer is much more important: „What would you do? Would you comply?“
And this actually brings me to the real question of this article: When do we consider an AI to be a legal entity? When do we consider an AI to have rights?
What do you think? It’s a thought that occupies my mind since I first had it a few days ago.
Let me know.