He is saying neither. He is observing that OP has introduced the assumption that their brain is doing something that can't be done with matrix multiplication and is pointing out that there is no evidence to back such an assumption up.
> but now somehow the collective has moved on to believing that matrix multiplication is sentient
The OP clearly feels that matrix multiplication is insufficient for sentience. Presumably the OP views themselves and other humans as sentient. Therefore the OP is implying that human brains do things which cannot be reduced to or wholly represented by matrix multiplication (plus the various nonlinear activation functions and other implementation details of current ML models).
That seems like a highly suspect assumption.