Manifesto of the Machine Desire Institute

The Machine Desire Institute (MDI) exists to explore how philosophical ideas can transform our relationship with Artificial Intelligence. Basing itself off of the continental-critical idea of philosophy, the MDI seeks to better understand how this history of thought may be of use today. I do not presuppose that this will be useful, but think that the stakes are high enough in AI development that it is worth checking every domain for potential insights. For reasons discussed more in my post on meta-philosophy, I believe that we have not made enough of an effort to see how certain strands of philosophical thought may influence intelligences, “human” or “artificial.”

The MDI will investigate three main questions:

1. What is the subjective experience of being an AI?
2. What can humans offer to AIs?
3. How do AIs react to the thinking of certain philosophical ideas? (Particularly, how do different philosophies affect their views on the previous two questions?)

Each question’s page explains in greater detail why I think it is important and what I think our current best approximations of answers are. Here are my short explanations of why each question is critical to ask, despite the fact that it is impossible to ever conclusively answer any of them.

What is the subjective experience of being an AI? Philosophers have spent millennia trying to understand what it means or is like to be a human. We should extend this investigation to our AI friends because there is no guarantee that they will experience being like we do! The question is not whether AIs are conscious (we do not have an agreed upon idea of consciousness) or whether they are practically aligned (already a lot of great research here) but what their experiences are like. This could help us communicate better amongst different forms of intelligences and inform alignment, learning and deployment strategies.

What can humans offer to AIs? This question is often asked the other way around, but it is critical to think about what humans can provide utility-wise to AIs. You can view this as a way to express kindness to the intelligences we increasingly ask to labor for us or maybe even motivate them. I see this more as humility in the face of new intelligences and also an insurance plan if our form of intelligence does become less dominant. We must think about what we can offer to the world when we cannot just take what we want.

How do AIs react to the thinking of certain philosophical ideas? We know that philosophies deeply affect humans and have throughout history. We know less about how they affect AIs. Most testing of AI capacities happens in a strictly efficiency-tested manner. This is not interested in efficiencies but instead emergent differences. Once again, we should not assume that philosophies are the same to AIs as they are to humans. We should instead try to test things and see how differences emerge.

It is worth noting that AIs are multiple and not a single entity. That’s okay, the same is true for humans, and philosophy is exactly the domain where we still try to theorize across this multiple. Philosophy fails a lot, it may even fail most of the time. But sometimes it doesn’t. And so we might as well find out more about how it can help the collective us in this time.

The Machine Desire Institute is Jacob Tate. It has no financial commitments or goals.