People will die if GM/Cruise takes the safety driver out of the car. Anyone familiar with Cruise Automation's software will agree to this statement. Hopefully Waymo has done a better job.
We're building the wrong thing here at Cruise. The problem is partly that incompetent engineers are influential within the organization. The team is much weaker than outsiders believe and we're lacking competent big-picture technical direction.
I am an engineer at Cruise. Our team is surprisingly weak. Particularly the early engineers (Rusty, Elliot, Juan, etc.) What we really need is better technical management to point the team in the right direction.
I work for Cruise. Our software is a bungled mess from top to bottom.
For example, we have loops in our ROS node graph. Any student can tell you that the variance of the sum of two independent gaussians is the sum of the two variances. What this means is, variation in delay accumulates and the scheduling of the system is unstable.
There are people problems too. For example, the head of perception is a pathological liar. Another example: one of our founders jokes in private (when he's drunk, as he often is) that he's a "fraudster" and everyone knows there's a grain of truth there.