Will AGI be conscious?

To recognize what it feels like to feel?

To interact with time and space?

To understand that our lives have a beginning and an end?

It is not clear yet how we could determine if an AI could become conscious. The language models that have claimed consciousness have so far been dismissed as likely “hallucinating,” repeating random messages assembled from past data.

Theoretically, if we could train an AI on new data without any references to consciousness and still continue to observe in the AI signs of consciousness, we could say it is real. But training an AI on large quantities of data without the mention of consciousness seems almost impossible.

Whether the ability to be self-sentient develops naturally within complexity or is a magical ingredient inherent only to humans, ultimately right now, it does not make much difference:

We will still have to deal with the repercussions of AI’s ability to replicate and surpass most of what we can do, without yet even needing to be “conscious”.