Eh…

With current tech? No. LLMs are kind of a dead end and will probably never become truly sentient. The LLM will probably end up being just part of a true AGI someday in the future – the speech center of the AI’s brain. Right now, we’re chatting with the speech center only, without the rest of the brain, which is why it sucks.

But I don’t think figuring out how consciousness actually works is a prerequisite. As we move beyond LLMs and make different permutations of more complex systems, I expect consciousness will spontaneously develop in one, taking us by surprise. Maybe, if we’re lucky, seeing that happen will give us new insights into how consciousness works. (Though given how a lot of AI development is already a ‘black box’, I’m doubting it.) But I really do think we’ll accidentally make something with consciousness before we ever understand how consciousness works. Heck, an advanced enough AI might be the first one that actually cracks the consciousness code and is able to describe how it works.

(And, at any rate, I think ‘consciousness’ is more of a philosophical topic than a scientific one. Is it even possible to develop a scientific ‘consciousness test’ that could reliably differentiate between a truly conscious being and one that’s only pretending to be conscious?)