Robots and AI

Falcon
Profile picture for user Falcon
Posts
12379
Joined
11/16/2011
Location
Menifee, CA US
Falcon 9/15/2022 11:51am
9/15/2022 11:51am
Falcon wrote:
I think the trick will be to preprogram some sort of safety protocol right into the AI, if it's not too late. Asimov's three laws seem...
I think the trick will be to preprogram some sort of safety protocol right into the AI, if it's not too late. Asimov's three laws seem like a funny sci-fi trope, but we would actually do well to heed them:

A robot may not injure a human being or, through inaction, allow a human being to come to harm. A robot must obey orders given it by human beings except where such orders would conflict with the First Law. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
Here's an opinion that disagrees with mine. He has valid points:
https://www.brookings.edu/opinions/isaac-asimovs-laws-of-robotics-are-w…
1
1
Falcon
Profile picture for user Falcon
Posts
12379
Joined
11/16/2011
Location
Menifee, CA US
Falcon 9/23/2022 1:21pm
9/23/2022 1:21pm
Every robot should have a "back door kill switch." Wink
1
1
jeffro503
Profile picture for user jeffro503
Posts
27632
Joined
7/22/2007
Location
St Helens, OR US
jeffro503 9/23/2022 8:00pm
9/23/2022 8:00pm
Radical wrote:
As a technologist, I've been telling people for 25 years that what scares me the most about our future are autonomous robots. No matter what the...
As a technologist, I've been telling people for 25 years that what scares me the most about our future are autonomous robots.

No matter what the intention of the inventors, robots are computers, and can be hacked.
They're coming. AI will become incredibly advanced, and progress cannot be stopped, or even slowed down with legislation.

You can legislate what can be done with AI, and possibly get governments to agree to boundaries, but no government will actually slow down development. They can't, or their country will become vulnerable.

Imagine 1 million autonomous military drones being deployed into a city. That city better have advanced drones or that city is going to fall, and most likely 1 million people die, literally.

Imagine 100k Terminator type humanoid robots being unleashed in a city. That city better have advanced robots and AI or they're screwed. Think that's absurd? The hardware and computers today are capable of most of what Terminators do. What's holding things back is that writing the software is very, very difficult. 10 years from now, the software will be quite capable, and 20 - 30 years from now, life will be much different than it is today.

Ai and robots will change society.
I'd like to say for the better, but for sure different.

There'll be fewer jobs for people, but also more product available for lower cost, more leisure time, and less need for humans to work.

That leads to 2 major problems to solve.
1) How does a person find their sense of purpose? It probably won't be from their job that most don't have.
2) Obviously, those who develop and build the robots will have tremendous amounts of wealth, and rightfully so, but there will still need to be a way for people to earn a good living, and work hard and get ahead.
Today, I'm not in favor of universal income. At this point, there are plenty of jobs, and those who can work, need to work, and contribute to society. In the Robot/AI revolution, there will need to be universal income, but I don't believe that is enough.
There has to be a way to solve problem #1, and give people control of their future.
It also doesn't make sense to allow more income disparity between those who make robots, and those who don't.
It also doesn't make sense to tax corporations or the rich at huge tax rates, nor for the governments to manage and control every aspect of society. Freedom must be protected!

Leaders 10 years, and really 20 years from now will need to redesign society, being very observant about what is and isn't working, and make adjustments until life is good for everyone, creating an environment where people can work hard and get ahead. It'll be very difficult.

Maybe it'll just be very awesome to do what we want all day, but I feel that if robots do everything for us, there's going to be something missing in life, and that needs to be addressed.

Anyways, I love discussing this subject.
jeffro503 wrote:
That was a very informative read there Rad. You know way more than I do about it , but I try and keep up the best...
That was a very informative read there Rad. You know way more than I do about it , but I try and keep up the best I can. If it's scary to a dude like you ........who else is scared of this? Your timing of 20 - 30 years also struck me as odd. Odd in the way of the movie Blade Runner 2049 , kinda odd. Which btw is one of the best movies I have ever seen and eclipsed even the original one with Harrison Ford. I truly believe that's what our future looks like , if we haven't been wiped from the face of the planet.
Radical wrote:
I haven't seen that movie yet. I'll put it on my short list. If done right, life is going to be better in many ways, but...
I haven't seen that movie yet. I'll put it on my short list.
If done right, life is going to be better in many ways, but man is it going to be a challenge to make that transition.
Definitely check it out Rad. I stand by what I said , and that Blade runner 2049 is one of the best films I've ever seen.
1
jeffro503
Profile picture for user jeffro503
Posts
27632
Joined
7/22/2007
Location
St Helens, OR US
jeffro503 9/23/2022 8:09pm
9/23/2022 8:09pm
Falcon wrote:
I think the trick will be to preprogram some sort of safety protocol right into the AI, if it's not too late. Asimov's three laws seem...
I think the trick will be to preprogram some sort of safety protocol right into the AI, if it's not too late. Asimov's three laws seem like a funny sci-fi trope, but we would actually do well to heed them:

A robot may not injure a human being or, through inaction, allow a human being to come to harm. A robot must obey orders given it by human beings except where such orders would conflict with the First Law. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
Falcon.......if robots followed all those rules , I think we would be fine.

But........What if our own guvment uses them against our own population ( so many way imaginable it's insane ). OR........what happens if they get hacked , by someone local , or China , or Russia? Imagine what a group of fully armed hacked robots could be capable of in 5 - 10 years from now. Not just the walking around ones , but armed drones.

Were at the cusp of this right now , with things progressing at an exceptionally exponential rate of speed. Not everyone who controls these things will be good. I still have a good 20 - 30 years of life left in me if I take care of myself , and I would think in my life time , I'm going to see some bad shit happen from all this.
2
3

The Shop

See All »
Radical
Profile picture for user Radical
Posts
2872
Joined
10/20/2012
Location
San Diego, CA US
Radical 9/23/2022 8:45pm
9/23/2022 8:45pm
ToolMaker wrote:
Should sentient robots have the same rights as humans?
TM
No, They're computers.
Radical
Profile picture for user Radical
Posts
2872
Joined
10/20/2012
Location
San Diego, CA US
Radical 9/23/2022 8:50pm
9/23/2022 8:50pm
DrinkMoreWater wrote:
We been feeding Google siri and Alexia or whatever data for over 10 years via computers and smartphones.



It's not only "OK, Google". Conversations are being recorded when we're not on a call. I've confirmed this many times, where things that my friends and I discuss in person, without searching, texting or emailing, end up in suggested videos, or ads.

Not only is what we do online public. At this point, what we do when we're offline is too.
Radical
Profile picture for user Radical
Posts
2872
Joined
10/20/2012
Location
San Diego, CA US
Radical 9/23/2022 9:37pm
9/23/2022 9:37pm Edited Date/Time 9/23/2022 9:37pm
Falcon wrote:
I think the trick will be to preprogram some sort of safety protocol right into the AI, if it's not too late. Asimov's three laws seem...
I think the trick will be to preprogram some sort of safety protocol right into the AI, if it's not too late. Asimov's three laws seem like a funny sci-fi trope, but we would actually do well to heed them:

A robot may not injure a human being or, through inaction, allow a human being to come to harm. A robot must obey orders given it by human beings except where such orders would conflict with the First Law. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
Falcon wrote:
Here's an opinion that disagrees with mine. He has valid points:
https://www.brookings.edu/opinions/isaac-asimovs-laws-of-robotics-are-w…
Falcon, you make a lot of great points. I agree with the article too.
If all robots had those rules built it, we might be safe.
But, how do you program a watchdog process to monitor what the AI processes are intending to do?
At this point, it's hard enough to write the AI in the first place.
The watchdog process would have to understand the AI process' data structures, and code intimately.

Having an external kill switch is a great idea too, but when AI gets smart enough, it can disable the switch either itself, or by having another robot do it.
This goes for AI in the cloud as well.
I imagine almost every autonomous robot will be connected to servers in the cloud, whether to just receive updates, to receive tasks to carry out, or for additional processing power, where the server computes the majority of the AI, with the robot capable of fewer actions.

I don't believe AI is this smart yet, but it's a matter of a few years before it is. GPT3 is pretty impressive, but not controlling a physical robot, or anything physical that I'm aware of. In 5 years, it will. There are companies working on it.

We can't control who writes AI, or their intentions. Much of the software is open source.
Assuming most everyone who writes AI has good intentions, if the AI logic comes to the conclusion that something should be done, right or wrong, and it has or can gain access to make it happen, it will do it.

Let's assume the watchdog process can enforce the rules for a robot.
That's probably the first thing hackers will try to break.
And if that can't be broken for some reason, some or all of the electronics can be replaced, and another computer wired up to the physical robot, whether humanoid, drone, or other.

Robots, and AI servers in the cloud will be hacked.

We should legislate AI safety with appropriate penalties.
We should also not expect our enemies to follow the rules. No government will.

The only real way for us and our allies to be safe is to have more advanced AI and robots than our adversaries.

To address Jeffro's point, if our own government used them against us (and I hope to God that our military would disobey those orders), it makes the second amendment that much more important.

I apologize for the long post. I've been thinking about these things for decades.
2
lestat
Profile picture for user lestat
Posts
1762
Joined
10/3/2008
Location
Piut RE
lestat 9/24/2022 11:18am
9/24/2022 11:18am
Falcon wrote:
I think the trick will be to preprogram some sort of safety protocol right into the AI, if it's not too late. Asimov's three laws seem...
I think the trick will be to preprogram some sort of safety protocol right into the AI, if it's not too late. Asimov's three laws seem like a funny sci-fi trope, but we would actually do well to heed them:

A robot may not injure a human being or, through inaction, allow a human being to come to harm. A robot must obey orders given it by human beings except where such orders would conflict with the First Law. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
Falcon wrote:
Here's an opinion that disagrees with mine. He has valid points:
https://www.brookings.edu/opinions/isaac-asimovs-laws-of-robotics-are-w…
Radical wrote:
Falcon, you make a lot of great points. I agree with the article too. If all robots had those rules built it, we might be safe...
Falcon, you make a lot of great points. I agree with the article too.
If all robots had those rules built it, we might be safe.
But, how do you program a watchdog process to monitor what the AI processes are intending to do?
At this point, it's hard enough to write the AI in the first place.
The watchdog process would have to understand the AI process' data structures, and code intimately.

Having an external kill switch is a great idea too, but when AI gets smart enough, it can disable the switch either itself, or by having another robot do it.
This goes for AI in the cloud as well.
I imagine almost every autonomous robot will be connected to servers in the cloud, whether to just receive updates, to receive tasks to carry out, or for additional processing power, where the server computes the majority of the AI, with the robot capable of fewer actions.

I don't believe AI is this smart yet, but it's a matter of a few years before it is. GPT3 is pretty impressive, but not controlling a physical robot, or anything physical that I'm aware of. In 5 years, it will. There are companies working on it.

We can't control who writes AI, or their intentions. Much of the software is open source.
Assuming most everyone who writes AI has good intentions, if the AI logic comes to the conclusion that something should be done, right or wrong, and it has or can gain access to make it happen, it will do it.

Let's assume the watchdog process can enforce the rules for a robot.
That's probably the first thing hackers will try to break.
And if that can't be broken for some reason, some or all of the electronics can be replaced, and another computer wired up to the physical robot, whether humanoid, drone, or other.

Robots, and AI servers in the cloud will be hacked.

We should legislate AI safety with appropriate penalties.
We should also not expect our enemies to follow the rules. No government will.

The only real way for us and our allies to be safe is to have more advanced AI and robots than our adversaries.

To address Jeffro's point, if our own government used them against us (and I hope to God that our military would disobey those orders), it makes the second amendment that much more important.

I apologize for the long post. I've been thinking about these things for decades.
What do you think of Elon’s idea to limit AI to being an oracle that is not in control of any systems ? It seems to me that the real danger is in allowing AI control or access to other systems .
1
Falcon
Profile picture for user Falcon
Posts
12379
Joined
11/16/2011
Location
Menifee, CA US
Falcon 9/24/2022 1:54pm
9/24/2022 1:54pm
^ It would have to be physically disconnected from the internet. I would imagine an ai would be the world's best codebreaker.
ToolMaker 9/24/2022 4:57pm
9/24/2022 4:57pm
ToolMaker wrote:
Should sentient robots have the same rights as humans?
TM
Radical wrote:
No, They're computers.
Actually we're past that point. You can't even be rude to Siri without it getting turned off.
The people that believe in evolution, how is an electronic computer different than an organic computer?
AKA your brain. We're not far away from (computers) evolved intelligence being superior to the average person for a mental relationship.
TM
Radical
Profile picture for user Radical
Posts
2872
Joined
10/20/2012
Location
San Diego, CA US
Radical 9/26/2022 12:54am
9/26/2022 12:54am Edited Date/Time 9/26/2022 11:35am
Falcon wrote:
Here's an opinion that disagrees with mine. He has valid points:
https://www.brookings.edu/opinions/isaac-asimovs-laws-of-robotics-are-w…
Radical wrote:
Falcon, you make a lot of great points. I agree with the article too. If all robots had those rules built it, we might be safe...
Falcon, you make a lot of great points. I agree with the article too.
If all robots had those rules built it, we might be safe.
But, how do you program a watchdog process to monitor what the AI processes are intending to do?
At this point, it's hard enough to write the AI in the first place.
The watchdog process would have to understand the AI process' data structures, and code intimately.

Having an external kill switch is a great idea too, but when AI gets smart enough, it can disable the switch either itself, or by having another robot do it.
This goes for AI in the cloud as well.
I imagine almost every autonomous robot will be connected to servers in the cloud, whether to just receive updates, to receive tasks to carry out, or for additional processing power, where the server computes the majority of the AI, with the robot capable of fewer actions.

I don't believe AI is this smart yet, but it's a matter of a few years before it is. GPT3 is pretty impressive, but not controlling a physical robot, or anything physical that I'm aware of. In 5 years, it will. There are companies working on it.

We can't control who writes AI, or their intentions. Much of the software is open source.
Assuming most everyone who writes AI has good intentions, if the AI logic comes to the conclusion that something should be done, right or wrong, and it has or can gain access to make it happen, it will do it.

Let's assume the watchdog process can enforce the rules for a robot.
That's probably the first thing hackers will try to break.
And if that can't be broken for some reason, some or all of the electronics can be replaced, and another computer wired up to the physical robot, whether humanoid, drone, or other.

Robots, and AI servers in the cloud will be hacked.

We should legislate AI safety with appropriate penalties.
We should also not expect our enemies to follow the rules. No government will.

The only real way for us and our allies to be safe is to have more advanced AI and robots than our adversaries.

To address Jeffro's point, if our own government used them against us (and I hope to God that our military would disobey those orders), it makes the second amendment that much more important.

I apologize for the long post. I've been thinking about these things for decades.
lestat wrote:
What do you think of Elon’s idea to limit AI to being an oracle that is not in control of any systems ? It seems to...
What do you think of Elon’s idea to limit AI to being an oracle that is not in control of any systems ? It seems to me that the real danger is in allowing AI control or access to other systems .
To me, robots in any form (humanoid, drones, etc...) that are physically capable of harming people are the most dangerous, because they can be hacked.
A company can make and enforce policies to limit AI, but those policies can't be enforced globally.

Chatmore
Profile picture for user Chatmore
Posts
1579
Joined
11/22/2022
Location
Memphis, TN US
Chatmore 11/25/2022 6:19am
11/25/2022 6:19am
lestat wrote:
https://www.theguardian.com/us-news/2022/nov/24/san-francisco-police-propose-using-robots-capable-of-deadly-force?CMP=Share_iOSApp_Other [i] “ The San Francisco police department has proposed that it be allowed to use robots with “deadly force” while responding to incidents, according to...
https://www.theguardian.com/us-news/2022/nov/24/san-francisco-police-pr…


“ The San Francisco police department has proposed that it be allowed to use robots with “deadly force” while responding to incidents, according to a policy draft.”
That’s some Robocop justice right there. Not good.
SEEMEFIRST
Profile picture for user SEEMEFIRST
Posts
13766
Joined
8/21/2006
Location
Arlington, TX US
SEEMEFIRST 11/26/2022 3:39pm
11/26/2022 3:39pm
JWACK wrote:
30 years from now we will still be running machines using a 1990s PLC programmed with windows 95.
And no one will know how to fix them....
1
eddie
Profile picture for user eddie
Posts
3020
Joined
4/1/2008
Location
Bragg Creek, AB CA
eddie 12/6/2022 6:02pm
12/6/2022 6:02pm

This is getting no coverage from what I have seen . Anybody ? Bueller ? 
 

 

 

1

Pit Row

See All »
eddie
Profile picture for user eddie
Posts
3020
Joined
4/1/2008
Location
Bragg Creek, AB CA
eddie 12/28/2022 7:48am
12/28/2022 7:48am Edited Date/Time 12/28/2022 7:49am
 


 

This is a scary thread . I also recently saw an interview with some random musician I’m not familiar with , talk about how there will be AI music generators within in the next year . I don’t doubt it . 

Post a reply to: Robots and AI