Although I completely agree that DL isn’t the brain, I find your reasoning a bit superficial. There are always counterpoints if you are looking at the literal difference between a silicon implementation and a biology implementation. For example, I am sure you have seen some of Hinton’s recent [talks](https://www.youtube.com/watch?v=VIRCybGgHts) arguing that the brain performs some forms of back-prop.
To me the biggest differences between DL and the brain is that task solving systems/models developed in DL are not complex systems. The chaotic properties of the brain are fundamentally different from the randomness in these model (even those with probabilistic neurones), which I believe they play a crucial role of human creativity and consciousness. Of course, it is always possible that some self-organized criticality will eventually emerge if we keep experimenting larger and larger DL model (kind of like Minsky’s thesis in his Society of Mind).
Since I already sound like a crazy guy from the internet, I might just as well throw in there: DL isn’t the brain because [the brain might be a quantum machine](https://www.quantamagazine.org/20161102-quantum-neuroscience/)