Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
In this article, we figured out that the peak potential of Artificial Intelligence in terms of emotions is yet to be achieved. While AI can mimic human emotions to a certain degree, it is not yet fully functional to replicate the same.
The Two Faces of Tomorrow by James P Hogan is my favorite AI story so far and it is from 1979.
At what point does an AI actually "understand reality" and can an intelligence exist without curiosity? I think a lot of this "Evil AI" is human beings projecting.
Can an AI feel pain to be mistreated by parents? A chess playing AI may be able to beat any human but does it give a damn about winning? Is it bored if humans don't play?
We can undoubtedly make robot tanks that would kill humans but if it is really an AI that understands is it going to ask why? It is the somewhat below AI that nitwit humans set off that I worry about.
That's the concern. A netwit human may activate some powerful AI bot to 'think' for itself. Or, that might happen even by accident. Hawking wasn't concerned about AI having emotions but letting advanced AI run amok on it's own.
Advanced AI will have the ability to update, edit, and write new code for itself it we allow it. That might be the only way to get AI more intelligent, so it can solve complex problems. Letting it 'free' to learn on it's own means letting it access what it needs to access, and edit it's code, all to become smarter.
For example letting AI free to do what it needs to do, may solve what are mysteries today, like cures for disease, the origin of the universe, etc.
It's 'thinking' like a human, except without emotions. The only thing guiding it is the coded tasks and goals it's suppose to do. But what if a task requires the AI to update it's code on it's own? It's then running autonomously.
Here's a scenario where an AI bot has to make a like or death decision. Say we made an advanced AI robot is tasked to provide home security. All bots, even security bots, will be instructed to 1. protect it's owner and occupants, and 2. do no harm to any human.
Then what if an owner of an advanced AI robot has their life threatened by an intruder with a weapon, that is clearly out to kill the owner of the house?
Say the AI bot finds the intruder in the hallway. The intruder turns and aims a high powered weapon at the bot. What does the bot do? If the bot lets the intruder shoot the bot, the bot may become disabled. The bot has already identified the weapon and calculated it can be disabled by the weapon.
The bot calculates all possible options and outcomes in nano-seconds. It's calculations find the only way to disable the intruder before the intruder pulls the trigger and disables the bot, is to kill the intruder. After all, it's first command is to protect it's owner. Even if law #2 is violated. No emotions, just calculated decisions. However, it made a decision to violate rule 2 to protect it's owner. That's AI at work.
The concern is if AI like this is set 'free' and given some difficult task, it will have to decide what to do with humans that get in the way of it's goal. It may decided to recruit other bots. And then disable humans in their way. All the bots under instruction of the bot set 'free', which is just doing it's task. AI amok!
And machines and systems can go amok. Often by human error.
Ever heard of cosmic bit flips? It f'ked up an election overseas one time and helped a player speed run the fastest record in SM64 yet cannot be achieved today with a huge bounty prize to whomever can replicate this! https://youtu.be/AaZ_RSt0KP8 The Universe is Hostile to Computers Tiny particles from distant galaxies have caused plane accidents, election interference and game glitches.
Now combine that ten fold and the AI running our house/car,etc. The sad thing is they make modern cars complex on purpose so we will WANT the stupid AI. "Ohhhhhhhh humans are tooooo duuuuuuuuuuummmmmmb to drive a car by themseeeeeeeeeeelves! We need compuuuuuteeeeeeers to do it for thhhhhhheeeeeeem!'
Ever heard of cosmic bit flips? It f'ked up an election overseas one time and helped a player speed run the fastest record in SM64 yet cannot be achieved today with a huge bounty prize to whomever can replicate this! https://youtu.be/AaZ_RSt0KP8 The Universe is Hostile to Computers Tiny particles from distant galaxies have caused plane accidents, election interference and game glitches.
Now combine that ten fold and the AI running our house/car,etc. The sad thing is they make modern cars complex on purpose so we will WANT the stupid AI. "Ohhhhhhhh humans are tooooo duuuuuuuuuuummmmmmb to drive a car by themseeeeeeeeeeelves! We need compuuuuuteeeeeeers to do it for thhhhhhheeeeeeem!'
Actually people probably are too dumb to drive. The leading cause of auto accidents? Distracted driving. Then weather, intoxication, aggressive driving, etc.
There are 40,000 auto accident deaths every year. When AI cars are driver-less accidents will be 99% reduced. You will not drive your self-driving car. When you turn on your vehicle it will register on the AI network and report your vehicle condition, stats and driver info. If approved to go then you'll instruct it where to take you. It will drive you there.
The AI road network will provide real time road and driving conditions, traffic, etc and take the best route. You will ride in comfort doing whatever you want- sleep, work on laptop, etc. That's actually a good thing if you think about it. Driving a vehicle is a waste of time when you could be doing other things.
An auto thief will have little incentive to steal your car- it has no steering wheel. Or gas pedal. It will only function on the AI road network. And the AI road network will immediately detect if an unauthorized vehicle is on the road and disable it. It might even lock the doors and drive the unauthorized vehicle to a police station.
Sorry, but this is the likely scenario in the future. But maybe 40 years from now. ICE vehicles will then be banned except for designated off network areas, where ICE enthusiasts can drive their gas car with a steering wheel for fun. Out of harm's way from the public.
Since driver-less cars will be too expensive for most people to afford you can rent a driver-less car when you need to go somewhere. When you do so- you don't maintain any cost of upkeep and you are not required to own auto insurance. That's actually a good thing.
So have fun now driving your ICE because the incentives to own an ICE will keep disappearing. The gov is moving towards electric and AI. However, you can still drive an ICE if you want to own one as a hobby and tow it to designated ICE only area. Or, have fun driving a virtual ICE with your VR headset. It will look and feel very realistic.
The AI road network will provide real time road and driving conditions, traffic, etc and take the best route. You will ride in comfort doing whatever you want- sleep, work on laptop, etc. That's actually a good thing if you think about it. Driving a vehicle is a waste of time when you could be doing other things.
An auto thief will have little incentive to steal your car- it has no steering wheel. Or gas pedal. It will only function on the AI road network. And the AI road network will immediately detect if an unauthorized vehicle is on the road and disable it. It might even lock the doors and drive the unauthorized vehicle to a police station.
Sorry, but this is the likely scenario in the future. But maybe 40 years from now. ICE vehicles will then be banned except for designated off network areas, where ICE enthusiasts can drive their gas car with a steering wheel for fun. Out of harm's way from the public.
I'm not seeing any signs whatsoever of a centrally controlled AI system. There is no road network. Everything points to autonomous independent vehicles with no external control or monitoring. In the long run, they might develop a road network and switch the autonomous vehicles to externally controlled vehicles but that is 100 years away. It will take 50 just to get everything autonomous.
I look forward to driverless vehicles but it's going to bring frustrations and inconveniences too. People envision cars driving around much faster because electronic reactions are faster but I think it will be the opposite with cars driving slower than what we are used to. If something is blocking the road, the driverless car is likely to just sit there endlessly rather than simply driving onto the shoulder or grass to go around. How will the driverless vehicle respond to human "traffic cops" at special events, gravel or grass field parking, temporary detours around construction or accidents, etc.
I'm not seeing any signs whatsoever of a centrally controlled AI system. There is no road network. Everything points to autonomous independent vehicles with no external control or monitoring. In the long run, they might develop a road network and switch the autonomous vehicles to externally controlled vehicles but that is 100 years away. It will take 50 just to get everything autonomous.
I look forward to driverless vehicles but it's going to bring frustrations and inconveniences too. People envision cars driving around much faster because electronic reactions are faster but I think it will be the opposite with cars driving slower than what we are used to. If something is blocking the road, the driverless car is likely to just sit there endlessly rather than simply driving onto the shoulder or grass to go around. How will the driverless vehicle respond to human "traffic cops" at special events, gravel or grass field parking, temporary detours around construction or accidents, etc.
Yes autonomous will probably be the first way it works, but every vehicle will be required to connect to the AI network for road data, traffic speeds, vehicles nearby, etc or it will not be allowed to be on the network road. The AI will monitor road conditions, know which vehicles and persons are in all vehicles, at all times. You will get real time status updates on your journey. AI will know where you and your vehicle are.
I'm thinking autonomous might prove to have too many glitches and they change it to full network control.
Seems to be the long term ultimate plan if you connect the dots.
It is not, to me anyway, any rogue AI or direct threat from AI that concerns me. It is the long term effect on humans. The more tasks AI does, intellectual ones in particular, humans become dependent and our ability to think, create and act wanes or disappears. Why bother -- AI is in charge.
yabutt .... how many of the humans do you figure are actually doing much critical thinking now? How much STEM is done without puter assistance today already?
I... The more tasks AI does, intellectual ones in particular, humans become dependent and our ability to think, create and act wanes or disappears. Why bother -- AI is in charge.
We will become dependent on AI, that's for sure. Like we are now with digital calculators. We could always do math manually. Or use a calculator. Should we stop using calculators?
AI will become profoundly intelligent and find solutions to very complex problems that would take years for people to figure out. One example would be winning a military campaign. Given all the data points AI will know how to win a combat.
Advanced AI will find solutions we might never figure out. One might be the solution to making a long lasting and quick recharging battery.
The problem is, what do you mean by AI? There are different systems doing different jobs better than we can, but none of them are capable of taking over the world. From simulated annealing, fuzzy logic, bigram models to genetic algorithms.
The problem would lie with any system that mimics how the human brain works, such as neural networks (specifically back propagation systems).
I have taken the following section of a post to explain why.
Quote:
Originally Posted by james112
It's 'thinking' like a human, except without emotions.
Why without emotions? Think of taste or colors. Sodium and Chloride do not have a taste characteristic, yet we taste the two together as salt. Colors do not exist, it is just a frequency of EM, but we experience them. This is how our brain evolved to understand the external world. What if emotions are just the brains way to understand our internal state?
And if this is due to the structure of the human brain, why should a model based on how neurons work not experience taste, color or emotions?
Quote:
Originally Posted by james112
The only thing guiding it is the coded tasks and goals it's suppose to do. But what if a task requires the AI to update it's code on it's own? It's then running autonomously.
Here is a typical error many people make, the programed robot where the tasks are coded. This is the error Plantinga made with his tiger. You could make a robot with coded responses, but then the robot must follow it's programming. It can react to input in a programmed way, but it cannot learn. If it finds itself in a situation it has not encountered before, it must have some default behavior that is identical for every situation it has not encountered before.
Neural networks do not behave like this. The program is is not telling the robot what to do (turn left, turn right, reverse), the code is telling the brain of the robot to work like our brain and learn from that input. This often allows it to approximate behavior in a situation it has not encountered before.
This also means a robot is able to learn from input we have not given it, and it is possible for a robot to learn something we have no control over, or to discover something we have overlooked.
Fortunately the networks we usually use are small. I often work with less than 20 neurons, and the network of networks used for facial recognition are still small compared to a brain. We also have the problem that the brain works in parallel, where as a neural network is both 1) serial, and 2) every extra neuron added increases the power required exponentially.
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.
Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.
Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
