How can the electrified tofu inside my skull give rise to the distinctive experience of tasting a fine red wine? How indeed can any physical process at all give rise to any kind of conscious experience at all? The long-term goal of my research is to understand the nature and physical basis of consciousness. Artificial intelligence has always played a role in this quest, and never more so than now. There are powerful synergies between the operation of artificial neural networks – the basis of most modern AI systems – and how brains work. One major opportunity I see is in ‘computational phenomenology’. Here, artificial neural networks inspired by theories of brain function are used to model properties of conscious experience. For example, we’ve been using deep generative networks based on theories of the brain as a ‘prediction machine’ to understand how visual hallucinations might arise. In the other direction, we can take inspiration from the brain to develop more powerful machine learning algorithms, closing the gap on things like energy usage and training data efficiency.

One of my main concerns is the persistent confusion over whether AI is, or will become, conscious itself. I think this is very unlikely. Brains are very different from computers, and intelligence is not the same thing as consciousness. For me, consciousness is more likely a property of life than of computation. Computational processes like AI can simulate the neural mechanisms of consciousness, but simulation is not the same thing as instantiation. It is important to recognise that major novel ethical considerations still arise from AI that merely seems to be conscious even if, under the hood, there is only subjective oblivion.