I’m not comfortable with giving my life into its hands without a spotless driving record for a year.
These questions are good:
“Personally, I’m not quite as concerned over the ability of the technology to ferry us around as I am with the underlying artificial intelligence involved. It sounds as if they are getting to the point where the cars can safely navigate their way around under normal circumstances. But when the unexpected happens (as it always will), I’d want to know a lot more about the “decision making” process going on under the hood. If three people suddenly leap out in front of my car with no safe way to avoid them and I’m the only passenger inside, will the car choose to hurl me off of a cliff to spare three lives at the cost of one?”
I definitely want flow charts and decision protocols before I buy one.
“Also, how much of the car’s intelligence is completely internal and how much (if any) is it relying on satellite data? If that GPS signal and other information are suddenly lost, can the car keep driving safely or will it just drift into the oncoming lane of traffic while I’m busy writing an article in the back seat?”
Excellent question! I assume all the ‘knowledge’ is locally cached and NOTHING is dependent upon a network which can fail at any time. I want guarantees from the manufacturer about this and a clear statement of their liability for all accidents they cause.
May not be “spotless” record but will soon (if not already surpassed) have substantially better records than human drivers. 10x safer will be very hard to say no to.
For Tesla at least:
All the AI learning is shared between all vehicles. They all learn from each other simultaneously. If one encounters a unique situation, it’s flagged and human trainers are involved to quantify & inform all vehicles about it. Each car is capable of driving entirely on its own, but certainly improves with network access.
Go look up Tesla Autonomy Day on YouTube. It’s worth the 4 hour watch to address everyone’s concerns.