While LaMDA almost certainly doesn't have a conscience, this argument doesn't really hold much water. Same can be said for the brain's of animals (why would neurons firing on a human brain create conciousness, but not in a worm?), or brains of people in irrecoverable comas. The computation being performed is what consciousness is, not the underlying hardware.

Your argument assumes that a machine simulating a concious is impossible. But if a perfect replica of your brain was modeled, and was simulated by giving it input at the same level of realism of your senses, according to your argument that simulation wouldn't be councious. But if you asked it, it would give the exact same answers that you would give to all questions, and in fact it's mind would be indistinguishable from your own.