Cardio
371
They are competent (in most situations) but they are not capable (especially in under-represented yet potentially high-stakes scenarios).
This would be how I characterize these AVs at the moment. The SF fire department once put into public sphere their nearly entire records of near-misses and bad misses with AVs. There was just another recent case that grounded the entire fleet away from freeways. These instances are surely not going to be the last because of the distinction I made, which is predicated on how machine learning occurs for this autonomous technologies.
Interestingly, the most insightful remarks I have encountered do not come from software and hardware developers but from philosophers like H. Dreyfus. To recapitulate this differently, you can't program the model of the world with all its requisite common sense and phenomenological intelligence into the machine--and this is simply a logical impossibility of nestled worlds. These cars lack common sense, which at times is dangerous, but for the most part, necessarily redeeming. And because they lack common sense (or what we call common sense as a driver), even quotidian everyday situations can become paradoxically dangerous--even if this is not perceived by the passenger as a near-miss. The hardware is also fitted with a remarkable array of sensitive and expensive sensors, as substitutes of the sensory faculties of the human driver; all which are counter-realistic to the rapid wear-and-tear of the urban road conditions, or angry and careless and irrational people on the road!
Despite this limitation, I still root for the development of this technology because it will change lives and mobility, especially people who are unable to drive due to physical or visual impairment. NYT recently published an article that describes the ethical possibility already known in this circle of technology. I think this is a good technology; but how to make the technological leap from what we have is another dimensional jump altogether.
There are a number of basic everyday scenarios and separate critical scenarios that make me question how autonomous vehicles can be approved beyond well defined areas and functions.
Ultimately I can’t blame the companies who are trying to develop these technologies because there’s crazy money to be had. However, it is the NHTSA that I believe has regulatory oversight and the ability to impose limits. A simple scenario is one where th...
Absolutely. Last year I used Waymo for an entire month....
And, like this article pointed out, despite being at the forefront of Self-Driving technologies, it still had certain issues that boggled my mind. Now, have certain elements iterated into considerably improved ones over the past year, most assuredly, but ...
Typical way how many US companies do business. Oversell to the point of lying.
I remember sitting in meetings where our team was presenting a benchmarking study of our software and how it stacks up against competitors. The most often asked question about our competitors was: „Is this a real capability or is it as advertised?“
They are competent (in most situations) but they are not capable (especially in under-represented yet potentially high-stakes scenarios).
This would be how I characterize these AVs at the moment. The SF fire department once put into public sphere their nearly entire records of near-misses and bad misses with AVs. There was just another recent case that grounded the entire fleet away from fr...
Imagine this in Paris?!?
A very interesting article from Euronews: www.euronews.com s The key discovery is normative drift: That AI agent actually follows the 'norms' dictated by the interactions of other AI agents. In other words, if the systemic interactions are general...
The industry calls these issues -long tail”
They are last -toughest, rare circumstances that come up in self driving. My son - is convinced our next vehicle will have no steering wheel! Although I think FSD is terrific; I’m not ready to give up the last human control.