@nafmo More resilient I'd say, it's designed for data integrity. I've had unrecoverable blocks on disks that damaged data and had zfs just automatically repair that data (this is in a simple two disk mirror).
I've also had memory go bad, had long running disk corruption as a result and zfs apparently still maintained integrity. Though in this latter case it was just backups so I did redo the backups from fresh just in case.
Have switched all three of my mdadm/ext4 pools over now. Yeah, I'm no power user. I haven't gotten into all the fancier zfs stuff like parity disks and snapshots yet.
I really like the zfs automated scrubs and associated alerts (via zed daemon), it feels like it's keeping watch on my data. With mdadm it always felt like I found out something was wrong when i felt other symptoms, like a corrupted file crashing an app. Maybe I could have been using mdadm better? I don't remember it doing anything like that.