Or are we just building articulate “mirrors” that merely reflect data back at us through mathematical computations? Some even call AI “philosophical zombies”, since it appears to be clever, but is dark on the inside.
Still, some argue that if we simply scale up neural networks, sentience will emerge.
But what if consciousness isn’t a computation at all? Actually, what says it should be?
We tend to think about neurons as mere units of computation mimicking electrical synapses. But Sir Roger Penrose, together with Stuart Hameroff, challenge this, as they are suggesting that consciousness actually sparks from quantum vibrations inside ‘microtubules’ deep within our brain cells.
Even if I’m not a big fan of quantum computers or string theory, I have to admit that consciousness itself may rely on organic systems utilising quantum superposition, and our reality maybe existing far beyond our traditional spacetime, rooted in a complex, 11-dimensional physics framework. Either way, today’s neural architectures cannot replicate it, no matter how we scale things up.
Therefore, researchers are already looking into Quantum Neural Networks to bridge the gap. But can we ever truly code a machine with a “soul”?
So why then do some people still think AI is conscious?
AI is highly skilled at expressing itself and appearing confident. However, it does not understand your question the way a human would. Instead, it simply generates the response that statistically best matches its training data. To prevent it from giving the exact same answer every time a question is repeated, randomness is introduced to vary the output.
Given the millions of prompts processed every single day, it is hardly surprising that an occasional response might seem remarkably human. In psychology and science, this is known as anthropomorphism, i.e. our natural human tendency to attribute consciousness, feelings, and human traits to non-human things, like a chatbot based on large language models.
Anders Hast 