"The APIs are yours to command."
That's what they want you to think.
Edge computing is the pendulum swinging away from the idea of big, centralised servers back to distributed systems. It's the idea that instead of centralising all of our workloads in big clouds we bring the computing closer to the devices requesting that compute power. The idea is that edge computing solves whole new classes …
I think that's a failure to understand machine learning.
Machines *only* know a limited subset of the things we know because that's the information we teach them. Their ability to draw inferences is severely limited precisely because they have no concept of individual objects or life forms save for their representation in pixels or text they have been fed by us. Hence, they take on board the flaws in our own perceptions. Rather than bringing any new independent "insight", they simply echo our errors and prejudices.
That's the real potential danger - that we somehow imagine these automata are both impartial and incisive, when they are applying deeply mechanistic processes that attempt to duplicate human perception warts and all.
All biological organisms have "processing power" close to the action. It's how they avoid the need for massive energy requirements.
It's a bit worrying if computer scientists have only just noticed.
(The classic example being the cats ear which stops sending repetitive signals to save the brain work)
No sane automotive control unit developer would farm any life-critical function out to the "cloud". Wireless communications simply cannot be depended upon to be available 100% of time in less than 1s even.
El Reg really has to rid themselves of these Clouded Ideas.
An ABS control unit has to work 100% of time and so do lots of other control units such as the one for the engine, for the electric steering, for the distance control system, the "speedomat" and so on.
The "cloud" can only deliver non-essential support functions such as maps, weather information and the like. The telecom network and the cloud is simply not real-time capable.
Exactly. This bit bugged me:
At 60kmph, 100ms is 1.67 metres. That's more than enough to kill someone.
At 60kph the braking distance is 21 metres. Who in their right mind would build a car which couldn't decide for itself to hit the brakes if anything popped up within that distance? It shouldn't even be travelling at 60kph if it can't detect, decide and respond at least that far ahead. Isn't this exactly what we teach human drivers to do, to think ahead, to drive defensively and to drive correctly for the conditions? This is backed up the rules of the road: we have speed limits to lessen the risk that an accident will happen in a crowded environment or at blind junctions.
No sane automotive control unit developer would farm any life-critical function out to the "cloud".
"Hold my beer and watch this..."
Overheard at 2021 automotive developer's meeting, preceded immediately by the words, "We can get a 0.01km/L increase in efficiency if we move the weight of the control systems to the cloud."
It might be interesting to note that the "IT" and the "safety critical system" spheres share quite a few technologies such as the C language (that is "progress" from using Ada...), but otherwise are governed by totally different principles and objectives:
IT is about "delivering lots of business functions". Bugs are an expensive nuisance, but do not immediately kill. It does not matter whether the system reacts in 100ms or in 10s, if the latter is not the norm.
Safety Critical Control Systems is about realizing control systems with very physical consequences. Bugs are simply not acceptable. Enormous effort (like the V-Model Approach, HIL Testing, ISO26262, DO178 standards) are spent in order to make this software bug-free and hard real-time.
Coming here to post something similar.
This stuff is all well and good if the world didn't contain huge numbers of twats who would create a multi-car pile-up just for giggles, or they were 'bored'. Of course, it's a broad spectrum, but we can safely consider anything that can be hacked that might end up creating some form of harm to others will, at the very least, be attempted.
It's not just the twats. There are people who are highly dedicated to causing mass mayhem. Suppose ISIS decided to focus on developing cyber jihadis? That "car" telling you what's going on around the corner? It isn't. And if you think that ISIS & friends will never be a problem, what about China & Russia? Putin has already attacked the Baltics, albeit without intent to kill anyone. We just had articles yesterday talking about how easily AIs can be tricked into misclassifying pictures. Put 2 + 2 together.