Well, not exactly. In a fully autonomous vehicle, the answer to the problem is defined by its software, and both the passengers and the public at large are tied to the tracks. The only way to break the premise is to decide that autonomous vehicles shouldn't exist. This is the critics' point.
The vehicle, presented with such a conundrum, simply stops. It is a default to a state of safety that avoids the very choice needing to be made.
How fast is it going? What's the stopping distance? What collision risk does it create by stopping? While autonomous vehicles theoretically have faster reactions than humans, it's still easy to create a scenario -- icy pavement, blind curves, falling debris -- where it can't react fast enough.
The speed of a self driving car - unlike a human driver; unless they are exercising special caution - is directly modulated by risk conditions.
To the extent that they correctly evaluate those conditions.
Something that can be continually improved. Whereas human driver performance declines over time. www.sciencedirect.com/science/arti...
The key question is “why is the car industry interested in autonomous vehicles”. Manufacturers marketed cars as an extension of personal freedom and the thrill of control. But the reality is that most vehicle accidents are down to preventable driver error committees.parliament.uk/writtenevide....
I think this is itself a trolley problem, though. Autonomous vehicles seem to be mostly worse *now,* but might get safer over time with more investment, while traditional motor vehicles are safer *now* but are also mature tech where future gains are likely to be incremental.
And you're absolutely right to point out elsewhere (I think) that decisions about the use of motor vehicles in general carry huge trade-offs as well.
There are also ethical differences between a company making decisions for all drivers of their vehicles vs individuals making decisions for themselves.
Sure. But these are questions that need to be asked before autonomous vehicles are deployed at scale, and saying "there is no trolley problem (for autonomous vehicles)" is an attempt to handwave the questions away. 1/2
The problems of deploying any technology at scale are usually under considered including the combustion engine itself.
The real world is not a trolley problem, but it is messy and unpredictable in a way software doesn't necessarily handle well. 2/2
Certainly agree with that
Or, as Qntm wrote a novel about, what if the autonomous vehicle encounters the chief programmer of a rival car maker?