Why do people think "AI" is necessary for simple things? Ordinary algorithms can adjust traffic lights based on traffic patterns.
Because 95% of people don't understand what AI in its current form is, yet they are convinced they do because they saw some movies about AI, or read some insanely stupid tweet from Musk.
As I sat at a light this morning for 90 seconds with no cars in sight, I finally reached my breaking point and decided to actually start this thread which I've been meaning to start for five years.
Can we stop talking about robotic cars and just start with a god damn smart traffic light that knows you are the only car within sight and lets you go?
Some dude is on another thread claiming a computer now can do AGI, well then why in the hell is the light red with no cars in sight?
You and me both on the topic of stoplights. Such a waste of life here in Colorado.
We are still a few human milestones away from reliable smart traveling vehicles.
Do you guys actually drive? These sensors are on probably less than 10% of lights - and hardly any of the major intersections which have the longest lights - and they are stupid as hell.
If I'm the only car at a major intersection and am in the left turn lane, I do not want to sit there for 2 minutes particularly when the only red light is a red arrow. The red arrow is the most infuriating light of al. Let me be the god damn judge of whether I can turn left or not at an empty intersection.
As for fully autonomous cars, here is my big worry. Let's say the tech worked perfectl 100% of the time. If we ignore pedestrians and animals, there would be no need for lights or even street directions. And self driving would naturally be made actually illegal.
And that's what scares me. The day they take the steering wheel out of the car is the day our freeedoms are really gone.
You’ve moved the goalposts from the smart lights need to be invented to they need to be more widely used.
"You’ve moved the goalposts from the smart lights need to be invented to they need to be more widely used."
or perhaps used better with perhaps more or smarter sensors on the main road to detect the traffic a bit further back
Yeah in the matter of a few sentences Rojo goes from upset that he had to wait at a light to fearing for his life that it will be illegal to drive cars. What the fvck
There is a magnetic strip embedded in the pavement a few feet (3?) behind the white stop line at a traffic light. If you stop well behind that line (a car length) or stop with your entire car ahead of the white line, the magnetic sensor can't "see" your car and the light won't change.
Motorcycles and scooters often aren't heavy enough to trigger these magnets. When I rode a motorcycle (Kawasaki Versys 650), it never triggered the sensor so I carried a copy of state law in my jacket that stated that motorcycles could run a red light if it didn't change.
Before this thread turns into the Montel Williams show, I think there needs to be protocols for what intelligence, smart technology, and AI mean.
Intelligence can mean common sense. You can make a right turn and make a U-turn. For smart technology, intelligence means the best solution for a given situation. The Wikipedia website says smart/intelligent, but expands on environmental impacts and other variables, not AI or human intelligence. One post confuses a garage door opener with AI. Whether a remote control uses a cord, Bluetooth, or microwaves it is not considering AI, but perhaps a smart solution.
Currently, AI is not the best solution to this problem, which is actually three different tasks. Those are:
1) identifying objects: AI --> deep learning --> computer vision is not at fifth grade level. That is why companies use sensors, cameras, and radar.
2) making a decision: This is rules based AI and there are an infinite number of rules that need programming.
3) transferring data to the processor: Where sensors and cameras can use edge computing, AI uses the cloud and experiences latency. Fog computing (dedicated local servers for self-driving cars) is in the process of development.
As I sat at a light this morning for 90 seconds with no cars in sight, I finally reached my breaking point and decided to actually start this thread which I've been meaning to start for five years.
Can we stop talking about robotic cars and just start with a god damn smart traffic light that knows you are the only car within sight and lets you go?
Some dude is on another thread claiming a computer now can do AGI, well then why in the hell is the light red with no cars in sight?
I recognize that this is not in the spirit of LetsRun, but have you considered calling your local traffic authority and lodging a request that the trigger mechanism be fixed if it is broken, installed at this location if it does not exist, or set to a shorter time interval if a sensor light is cost prohibitive?
Be part of making the world a better place instead of acting like a spoiled child.
The problem is that there are lots of places that don't have a "pattern." Think of a scenario where there is basically no traffic turning left through multiple busy lanes onto a tiny side street. But once the light is put in, they either have to set up a sensor, or stop the major through-street every minute to let a car turn left (if there even is one). If you lived in a town without the budget to put in sensors, then your main road just got slower (stopping every minute for no reason) or you strand people at the small street waiting forever for it to turn green.
The sensors are for lights on side streets that have very little traffic and have a long delay. You have to drive up onto the sensor to get the light to change. Otherwise, the light will stay red for the side street. Rojo wants to have a sensor that detects a lack of traffic on the busy street and shorten the cycle for that street in favor of the less traveled side street.
That kind of technology is really not out there, but is part of what some will hope to be infrastructure for future autonomous vehicles. Truly autonomous vehicles (i.e. driver can fall asleep in the back seat and auto pilot does everything) will need to be able to communicate with each other and with traffic signals. The idea is to be able to seamlessly move autonomous vehicles through intersections without having to have one direction stop and wait while the other direction has a green light. Vehicles would see each other coming and slow down or speed up just enough to be able to go through the intersection without stopping. That would eliminate vehicle idling and save a lot of energy while reducing commute times. That is the real future AI sh#t (and is probably a few decades away despite AI proponents' predictions that everything will be possible in 24 months).
This is terrifying to me. Imagine there’s a bug in the software…
I recognize that this is not in the spirit of LetsRun, but have you considered calling your local traffic authority and lodging a request that the trigger mechanism be fixed if it is broken, installed at this location if it does not exist, or set to a shorter time interval if a sensor light is cost prohibitive?
Be part of making the world a better place instead of acting like a spoiled child.
Don’t be such a prick.
Actually contacting the city engineering dept. if there is an actual problem, is the right action. This type of problem (6 or 7 minutes, not 90 seconds) actually happened on Duke's East Campus. Enough people complained that the city investigated the issue and subsequently returned the fines and had the tickets taken of people's driving record because they determined it was a problem with the traffic light.
However, personally I would not get my boxers in a wad over just 90 seconds.
AGI is learning how to do government functions by learning from government websites and laws. As such, it will behave with government efficiency.
Using AI to amplify human intelligence is a great topic, but you have managed to turn it into one of the stupidest posts in the history of LetsRun (and that is quite an accomplishment). There is no old AGI, so I'm not sure how you can have new AGI. AGI is not learning, AI is learning. There is a leap of faith between the two sentences since AI does not have reasoning capabilities. Even if AI figured out how to run the country like a company, which is essentially what Trump/Musk are planning, that is not AGI since it is performing a single task.