My name is Nora Belrose. I’m an AI researcher who’s spent the last few years thinking deeply and reading widely about the philosophical implications of the AI revolution we’re going through right now. I think there are four big ones:

1. The mind is not a mechanism.
   Artificial intelligence started working when researchers gave up on the idea that intelligence could be programmed like normal software, and started allowing computers to learn on their own using a technique called deep learning. While traditional software is made of discrete symbols, modern AIs are holistic networks of continuous variables which have no meaning individually.
2. The world is not a mechanism either.
   Deep learning is starting to replace explicit scientific theories, performing tasks like weather forecasting and protein structure prediction more accurately and efficiently than any theory-based simulation. This means that theory is just one of many ways to understand the world. We cannot reduce reality to any theory.
3. Computers cannot feel. Feeling is flow, computation is control.
   AI is getting better and better at faking photos and videos, and it’s quite good at faking feelings, too. Despite the continuous and holistic nature of AI software, today’s AIs still run on digital hardware. This sharp division between hardware and software allows us to copy, edit, pause, and repeat AI programs at will— it’s what we use to control them and shape their behavior. It’s also precisely what makes them incapable of genuine feelings. Consciousness is an irreversible, untamable, one-of-a-kind flow of feeling, emotion, and meaning, not a controlled computation.
4. Ethics is about unleashing creativity and love.
   Feelings are central to ethics, but that doesn’t mean we should try to minimize pain and maximize pleasure. Since feelings are flows, not computations, they can’t be copied, counted, or numerically compared across individuals. Instead, ethics is about empowering people to create, explore, and love.