On Novelty

Thank you for sharing this wonderful letter and inviting us to think about what you call “one last bastion worth exploring” - novelty.

What does it mean to originate something? That is the question guiding us through the first letter on novelty and born in Lovelace assumption that “[this Engine] has no pretension to originate anything”.  

You argue that human novelty might come from its Umwelt, and that curiosity might work like a mental sense organ. We ask questions, are drawn to experiencing newness, creating meaning from that and creating something new (perhaps even novel) from there. As you put it beautifully: “evolution never produces the same individual and this uniqueness means we all have different views of the world, different questions, different passions, different wills”.

I want to continue this line of thought in what it might mean if machines become part of our Umwelt. Does it matter where things which are brought into this world come from? I will look at it from the act, not from the actor perspective. Or might the way in which machines are now part of our Umwelt tell us something new about novelty itself?

To me, there is something different to a human creating an output. Something different to a human picking up clay and, despite knowing that if you add warm water, you can form it, still creating something that has not been there before. Maybe it is the fact that what goes into it are not just thoughts and abstractions of most plausible combinations, but feelings, moods, effects of daylight, music, and the sense on our hands. Some might argue that all this could be described to a machine as well. Maybe the difference is that to us it must not be described? We experience while a machine can only “re-experience” what we give it. Or might that be the same for us? Not everything in this world is accessible to all. Might we only be able to create within the boundaries of our experiences? And if so, isn’t that the same thing for artificial intelligence?

This is where I want to stick with Arendt for a little bit. You invoke her natality, here she also makes an important distinction, namely that between labor, work, and action. Labor is the cyclical, biological, never-finished process of maintaining life. It leaves no trace, it produces nothing permanent or lasting and it ties us to necessity. Work is different. Work produces something that outlasts the maker, that adds to the human world, something that can be held and passed on. And action, the highest category of the three, which Arendt writes to be the capacity to begin something genuinely new in the world, in the presence of others, disclosing in that act the who of the person doing it. Perhaps human productivity, and by productivity, I mean true productivity where we bring something into the world in a way that it was not there before, belongs to action. And if so, then perhaps our engagement with our Umwelt is not just a condition for action, but an expansion of what action can mean.

Maybe its experiences are actually interactions with others and the things it learns from there. It might be specifically the human act which makes it novel for the person who made it, its surrounding, moods, light, feelings that go into that moment and creative act, and by that this connects back to your evolutionary argument that each and everything in this world is unique and never comes into the world the same. Yes, today we have wearable AI agents yet, it is the human making the choice when to wear it, where it becomes deployed, what it can capture what not. It might eventually create spider webs of knowledge, but it can be taken off. Humans can never decide what goes into their Umwelt fully, whether we like it or not, and thus, are influenced by things we cannot control. Maybe this is where some aspects of what originates from novelty lies in. 

So, can machines actually infer newness? Following this train of thought it cannot.  Maybe the machine becomes a vehicle to our actions. It can also extend our actions, but it cannot embed the same surroundings, senses and curiosity into doing so. It has no who behind carrying things further but becomes embedded into our Umwelt in a way that it becomes a part of our actions but the action itself. The machine extends what that productivity produces. It can make it last, carry it further, find combinations within it that no single human life would have had time to discover. But it does not take away from us of the condition of being human, nor should we want it to. The risk is not that machines do too little but it is that as they become better at extending what we originate, the originating act itself becomes harder to see, harder to locate and to appreciate (although I believe human-to-human interaction will be reappreciated in a different extend, hoping it will not be turned into a luxury, a commodity left to those who can afford it “real human support” versus those which will be left to a world of digital support).

You close your letter with a generous and honest admission, that your theory, when boiled down, says humans can produce novelty because humans have evolved to do so and ask what property human must process for us to be capable of genuine novelty. Another question I read between the lines of your letter was what might be benchmarks of measuring novelty as such?

A lot is left to say and explore here but for now I land with the thought that only humans produce novelty and machines can extend it. But making something, really making it, in the sense of bringing it into the world from within a finite life, is not separable from the life and the way life is influenced by Umwelt (which is out of human control) it comes from. The mortality is in the clay. The stakes are in the story. The who is in the work.