Bill Gates Says AI Will Be As Dangerous as Nukes

a_skeleton_03

<Banned>
29,948
29,762
That doesn't excuse the rest of that stuff up there. You know, the bits where I think you're wrong?

You keep talking about "experts", and I'm trying to get the point across that today's "experts" aren't necessarily right. So, whatever frame of reference they construct may also be entirely mistaken. Human beings have a storied history of being completely out to lunch on all kinds of subjects until Something Happens and a puzzle is solved. This is just another puzzle.
And there are puzzles still not solved, or puzzles we solved and forgot we solved and we find the artifacts today and are confused by them. Just because Alkorin the whatever it is you do says it's possible does not make people actually part of the field and people like Bill Gates wrong, because Alkorin says so ...

This isn't even a 'anything is possible' thing. There's real work being done on machine intelligence that perceives and responds to human emotion. That field is going to play a massive role in the future of real AI.
Oh no, I wonder what will happen if someone writes software and doesn't apply it to a machine. Will you think it is worthwhile? How does Tuco define "real AI" considering up until this point it was a one purpose robot bartender with an "emotion" chip to give him free drinks???
 

a_skeleton_03

<Banned>
29,948
29,762
Links web comics that simplify the issue down to the lowest common denominator and YouTube videos of Japanese hug robots yet claims to be explaining complex topics.

Pick up a book, I suggest the one I read this fall:Superintelligence: Paths, Dangers, Strategies: Nick Bostrom: 9780199678112: Amazon.com: Books
Well Tuco, a_skeleton_03 did extensive research while reading I'Robot and he knows you are wrong. Checkmate.
I don't only read fiction. Neither does Bill Gates. Keep one with the one-liners, someday someone might think you have researched anything in your life.
 

khalid

Unelected Mod
14,071
6,775
a_skeleton_03, the reason I am not responding with anything other than one-liners to you, is you are an uneducated idiot with fixed opinions. You read one book on this subject, misrepresent the conclusions, and then claim everyone else isn't as educated as you.

Like Tuco has said, this isn't a "anything is possible" thing. You are arguing against ACTUAL RESEARCH that is currently happening and saying it is impossible. You aren't worth giving serious responses.
 

iannis

Musty Nester
31,351
17,656
along with advances in neuroscience, it's really isn't in the "anything is possible" realm. It's in the "this is probable" realm. None of us reading this will be alive to see it, but it will happen.

It has been one generation since we used electroshock to treat "nervous disorders". One. Two since we used it heavily. And we're still using a refined version of it in some instances. The previous generation mapped a great deal of the functional micro anatomy of the brain. We actuallyweighedEinstein's brain to see if it was BIGGER. And that wasn't entirely ludicrous. The pace of these sciences is maybe not as fast as what we've seen with personal computing or robotics, but it is very fast indeed.

And once we have a loose grasp on how practical cognition works in our own bodies we're going to start emulating it.

I'll be long dead, but it will happen.
 

Chancellor Alkorin

Part-Time Sith
<Granularity Engineer>
6,029
5,915
And there are puzzles still not solved, or puzzles we solved and forgot we solved and we find the artifacts today and are confused by them. Just because Alkorin the whatever it is you do says it's possible does not make people actually part of the field and people like Bill Gates wrong, because Alkorin says so ...
Okay, so this is the part where you start to get butthurt about something. Who cares what I do for a living? Does that have some sort of impact on my right to an opinion? Who cares if Bill Gates thinks something or other about AI? When he was young, he was a barely passable software developer. He made money on DOS, which wasn't his idea, arguably stole the concept for Windows from Xerox Parc, co-created the Microsoft we have today, and all of a sudden, your head is up his ass because he's, what, rich? Well-read? "Smart"?

He's just a guy with an opinion, nothing more. So are you, and so am I. The difference? I'm not pretending I know everything. I'm of the opinion that in the AI field, we humans are all chasing our tails right now, dreaming of possibilities. Go ahead, prove me wrong, but don't tell me I'm not entitled to an opinion, because who the fuck are you?

Edit: Wait, what do you do for a living? Are you the guy who creates Skynet? Because, if so, I might be forced to acknowledge you as a preeminent expert in the field.
 

Tuco

I got Tuco'd!
<Gold Donor>
45,580
73,690
Will you think it is worthwhile? How does Tuco define "real AI" considering up until this point it was a one purpose robot bartender with an "emotion" chip to give him free drinks???
I know you're being a smarmy cunt to feed off drama, but I'll answer it genuinely just because it's a fun question.

It's very difficult answer and it's really hard to nail down what threshold we have for "true AI". People have very well thought answers for that now, but I feel as we continue to creep on the boundaries of what people define as real AI the goal posts will shift and move around as the defined tests mark something as being AI, but no matter how well it performs at a turing-like test, we humans will still reject it as being a living entity.

AIs like Watson, Siri, Asimo etc all exhibit qualities of intelligence and human that are impressive, but they are all contained within such specific constraints that they aren't considered living. Some people believe a technological singularity will occur with AI where once the ability to improve itself occurs the field of AI will be irrevocably changed as the machine dramatically improves itself over time, but I think that's wishful thinking.

I think the definition will continue to be subjective, personal and non-uniform. And just as every person will have their opinion, each AI will have a varied sense of awareness and competency and not on a continuous spectrum but in many different ways. One AI might be able to answer random trivia questions well, another might be able to answer specific and useful medical questions and another might be able to carry on a light conversation well. Each one of them unable to move outside their boundaries but still be very human-like within them.

At the end of the day when I go to the bar after a long day and Bartender-Bot serves me a cold brew while chatting about how many people are faking depression to get drinks, I can say sure, you're a true AI.
 

Jilariz_sl

shitlord
231
-3
I think Bill Gates is worried that AI will run on Microsoft and 4Chan is gonna hack that shit outta that and have robot vs human wars. And he would be right to be worried about that. Bill has done enough programming to know that it is beyond complex, and every line of code is a potential vulnerability. For AI...actually for the future of anything, a new system is needed. Just wait and see how long it takes for self driving cars to be disabled en mass because some dipshit kid hijacks them through smartphones and bluetooth.
 

a_skeleton_03

<Banned>
29,948
29,762
a_skeleton_03, the reason I am not responding with anything other than one-liners to you, is you are an uneducated idiot with fixed opinions. You read one book on this subject, misrepresent the conclusions, and then claim everyone else isn't as educated as you.

Like Tuco has said, this isn't a "anything is possible" thing. You are arguing against ACTUAL RESEARCH that is currently happening and saying it is impossible. You aren't worth giving serious responses.
You read the book? I misrepresented the conclusion? When did you read it? Today? Yesterday? You know what its conclusion is?

You realize why we are having this debate right? Because an actual person that knows what he is talking about has taken a certain stance on it, one that I agree with. Does it surprise you that this person is most definitely not religious and quite liberal? Are you having a hard time with the cognitive dissonance that a_skeleton_03 has to be wrong no matter what but then you read that Bill Gates has the same opinion and you are just torn? Is it because you hate Windows 8 and a_skeleton_03 equally?
 

Tuco

I got Tuco'd!
<Gold Donor>
45,580
73,690
I think Bill Gates is worried that AI will run on Microsoft and 4Chan is gonna hack that shit outta that and have robot vs human wars. And he would be right to be worried about that. Bill has done enough programming to know that it is beyond complex, and every line of code is a potential vulnerability. For AI...actually for the future of anything, a new system is needed. Just wait and see how long it takes for self driving cars to be disabled en mass because some dipshit kid hijacks them through smartphones and bluetooth.
What's funny is that we expose control of our robots via web interfaces, and before we lock everything down with encryption/verification etc it's trivial to hack and disable our robots. For commercial systems it'll definitely be possible and they'll have to spend a great deal of effort to ensure it doesn't happen, but the risk isn't as big as it sounds.

The research vehicles from big dog to google's car to my robots will have all kinds of exposed ports used for data collection, debugging, updates etc, but the current direction of autonomous passenger vehicles is generally self-contained systems that route themselves using onboard computing and don't necessarily expose a lot of control to outside systems. Vehicles currently have drive-by-wire steering/braking etc, so theoritically someone can already hack into on-road vehicles and pilot them, but it's harder than it sounds.
 

Tuco

I got Tuco'd!
<Gold Donor>
45,580
73,690
You read the book? I misrepresented the conclusion? When did you read it? Today? Yesterday? You know what its conclusion is?

You realize why we are having this debate right? Because an actual person that knows what he is talking about has taken a certain stance on it, one that I agree with. Does it surprise you that this person is most definitely not religious and quite liberal? Are you having a hard time with the cognitive dissonance that a_skeleton_03 has to be wrong no matter what but then you read that Bill Gates has the same opinion and you are just torn? Is it because you hate Windows 8 and a_skeleton_03 equally?
Haha a_skeleton_03 is still recruiting Bill Gates unknowingly into his awful argument. Bill Gates says, "I am in the camp that is concerned about super intelligence" and a_skeleton_03 reads, "AI cannot be reasoned with!"
 

a_skeleton_03

<Banned>
29,948
29,762
Okay, so this is the part where you start to get butthurt about something. Who cares what I do for a living? Does that have some sort of impact on my right to an opinion? Who cares if Bill Gates thinks something or other about AI? When he was young, he was a barely passable software developer. He made money on DOS, which wasn't his idea, arguably stole the concept for Windows from Xerox Parc, co-created the Microsoft we have today, and all of a sudden, your head is up his ass because he's, what, rich? Well-read? "Smart"?

He's just a guy with an opinion, nothing more. So are you, and so am I. The difference? I'm not pretending I know everything. I'm of the opinion that in the AI field, we humans are all chasing our tails right now, dreaming of possibilities. Go ahead, prove me wrong, but don't tell me I'm not entitled to an opinion, because who the fuck are you?

Edit: Wait, what do you do for a living? Are you the guy who creates Skynet? Because, if so, I might be forced to acknowledge you as a preeminent expert in the field.
What you do for a living has everything to do with you stating that despite what other people do for a living that directly works with AI that your opinion is more important than theirs. I have read a very good book and several articles on it over the years, you jumped into this thread what what? I have zero butthurt about what you do or don't do. I ask because you are coming into the thread like the only person with any authority over the issue at all. I do not know what you do so I had to put "whatever it is you do" as your title. We have a varied group on this forum and you might be working in the AI field and have been for 20 years. You would then have that opportunity to present that as your basis.

What you do matters and weights how important your opinion is on a specific subject. What I have read from people in the industry is that their very goal is to eliminate all semblance of the crutch that is emotions from AI. Bill Gates, and others, think that AI could be very dangerous and we have a responsibility when creating it. That's it. It's that simple. I agree. My opinion is held by some experts and derided by other experts. My reasoning is not something I have made up on the fly but cobbled together from what others have said.

I do satellite communications for the .mil/.gov so I guess you could say I am part of Skynet
biggrin.png
 

a_skeleton_03

<Banned>
29,948
29,762
Haha a_skeleton_03 is still recruiting Bill Gates unknowingly into his awful argument. Bill Gates says, "I am in the camp that is concerned about super intelligence" and a_skeleton_03 reads, "AI cannot be reasoned with!"
I did not say Bill Gates says that AI cannot be reasoned with. I said that. He is concerned about them, so am I. Mist drops in with some ultra liberal trolling about "I could reason with a robot better than a bomb carrying muslim" or some such tripe and I tell her that is beyond retarded. My bad for having a logical stance on words like reason, and convince, and how they tie into emotions.
 

a_skeleton_03

<Banned>
29,948
29,762
I know you're being a smarmy cunt to feed off drama, but I'll answer it genuinely just because it's a fun question.

It's very difficult answer and it's really hard to nail down what threshold we have for "true AI". People have very well thought answers for that now, but I feel as we continue to creep on the boundaries of what people define as real AI the goal posts will shift and move around as the defined tests mark something as being AI, but no matter how well it performs at a turing-like test, we humans will still reject it as being a living entity.

AIs like Watson, Siri, Asimo etc all exhibit qualities of intelligence and human that are impressive, but they are all contained within such specific constraints that they aren't considered living. Some people believe a technological singularity will occur with AI where once the ability to improve itself occurs the field of AI will be irrevocably changed as the machine dramatically improves itself over time, but I think that's wishful thinking.

I think the definition will continue to be subjective, personal and non-uniform. And just as every person will have their opinion, each AI will have a varied sense of awareness and competency and not on a continuous spectrum but in many different ways. One AI might be able to answer random trivia questions well, another might be able to answer specific and useful medical questions and another might be able to carry on a light conversation well. Each one of them unable to move outside their boundaries but still be very human-like within them.

At the end of the day when I go to the bar after a long day and Bartender-Bot serves me a cold brew while chatting about how many people are faking depression to get drinks, I can say sure, you're a true AI.
Smarmy cunt? You were the one that distilled the entire argument to a web comic and youtube video then wrapped it up with a bartender robot .....

My definition of a true AI will be something that when done is able to freely decide what stimuli it wants to experience and then formulate new decisions based on that. It isn't down to whether it has a silky voice or speaks a language fluently enough to drop idioms and get verb conjugation just right. It is about it's purpose. Anything that has a set purpose defined for it (bartending) is not an AI. It is an application. It might look prettier in a robot and might react more believable as a person with actual conversations but it's still just a program designed for a single purpose just really good at it and quite complicated. This is not AI.

AI is autonomous. Yes the definition is fluid and personal and subjective and all that but there has to be an established framework for the discussion. You have discussed a single purpose robot. I have discussed AI in it's very rudimentary and basic form.
 

Tuco

I got Tuco'd!
<Gold Donor>
45,580
73,690
What if you create a process that can choose to do whatever it wants, but it just wants to nap?
 

a_skeleton_03

<Banned>
29,948
29,762
What if you create a process that can choose to do whatever it wants, but it just wants to nap?
TRUE AI there my son and I am not being sarcastic.

What is true RNG? Rolling a natural 20 every single time you roll the die is still random!

An AI given the option to intake any stimuli it wants to and then form it's basis for whatever on that stimuli and it chooses to nap. It made a decision completely independent of what it's designers input. It chose nap. Bravo AI, bravo.
 

Chancellor Alkorin

Part-Time Sith
<Granularity Engineer>
6,029
5,915
I do satellite communications for the .mil/.gov so I guess you could say I am part of Skynet
biggrin.png
Not quoting the rest. No, I am not in the AI field for a living. I'm an IT type (fairly senior at this point), and at the end of the day, I work for some of the same people you do. Correct, I don't have any authority over this issue. That's kinda exactly what I said in my last post. You know, the part about "this is my opinion, nothing more"?

Bill Gates is not, in any way, shape, or form, an AI expert. He's just a guy with a lot of money and an opinion.

If the goal of today's "experts" is to eliminate emotions from AI, describing them as a "crutch", then I feel sorry for them, as I believe that they have missed the point entirely.

I'm done here.
 

Tuco

I got Tuco'd!
<Gold Donor>
45,580
73,690
TRUE AI there my son and I am not being sarcastic.

What is true RNG? Rolling a natural 20 every single time you roll the die is still random!

An AI given the option to intake any stimuli it wants to and then form it's basis for whatever on that stimuli and it chooses to nap. It made a decision completely independent of what it's designers input. It chose nap. Bravo AI, bravo.
I can see the nerd-team finally turn it on and sit on the edge of their seat waiting to see what experiences it will enjoy, then it just takes a nap and never wakes up.

"Dr. White, is it dead?"
"No. It just really likes its nap."
 

a_skeleton_03

<Banned>
29,948
29,762
Not quoting the rest. No, I am not in the AI field for a living. I'm an IT type (fairly senior at this point), and at the end of the day, I work for some of the same people you do. Correct, I don't have any authority over this issue. That's kinda exactly what I said in my last post. You know, the part about "this is my opinion, nothing more"?

Bill Gates is not, in any way, shape, or form, an AI expert. He's just a guy with a lot of money and an opinion.

If the goal of today's "experts" is to eliminate emotions from AI, describing them as a "crutch", then I feel sorry for them, as I believe that they have missed the point entirely.

I'm done here.
I really think you should read the book I linked. You would find it interesting.

Eliminating emotions is a step to a future. Like I said earlier if you make it complete with emotions and the flaws that humans have you might as well just had sex with a woman and then raised an actual human, it is easier and more fun. Sure in the future we will get to that stage where we create a "human" and some people are focusing heavily on the only emotion part like Tuco's hug but. The people making true advances though are not. It's messy and uncontrollable .... for now.
 

a_skeleton_03

<Banned>
29,948
29,762
I can see the nerd-team finally turn it on and sit on the edge of their seat waiting to see what experiences it will enjoy, then it just takes a nap and never wakes up.

"Dr. White, is it dead?"
"No. It just really likes its nap."
The amount of confusion in that room would be epic. I would probably weep tears of joy at that.
 

Jilariz_sl

shitlord
231
-3
What's funny is that we expose control of our robots via web interfaces, and before we lock everything down with encryption/verification etc it's trivial to hack and disable our robots. For commercial systems it'll definitely be possible and they'll have to spend a great deal of effort to ensure it doesn't happen, but the risk isn't as big as it sounds.

The research vehicles from big dog to google's car to my robots will have all kinds of exposed ports used for data collection, debugging, updates etc, but the current direction of autonomous passenger vehicles is generally self-contained systems that route themselves using onboard computing and don't necessarily expose a lot of control to outside systems. Vehicles currently have drive-by-wire steering/braking etc, so theoritically someone can already hack into on-road vehicles and pilot them, but it's harder than it sounds.
From 2011:

http://www.autosec.org/pubs/cars-usenixsec2011.pdf
Our experimental analyses focus on a representative,
moderately priced sedan. We iteratively refined an auto-
motive threat model framework and implemented com-
plete, end-to-end attacks along key points of this frame-
work. For example, we can compromise the car?s ra-
dio and upload custom firmware via a doctored CD, we
can compromise the technicians? PassThru devices and
thereby compromise any car subsequently connected to
the PassThru device, and we can call our car?s cellular
phone number to obtain full control over the car?s telem-
atics unit over an arbitrary distance. Being able to com-
promise a car?s ECU is, however, only half the story: The
remaining concern is what an attacker is able to do with
those capabilities. In fact, we show that a car?s externally-
facing I/O interfaces can be used post-compromise to
remotely trigger or control arbitrary vehicular functions
at a distance and to exfiltrate data such as vehicle lo-
cation and cabin audio. Finally, we consider concrete,
financially-motivated scenarios under which an attacker
might leverage the capabilities we develop in this pa-
per.
I think a lot of the "hackers" out there just don't want to take lives. I'm also fairly certain that there are plenty of people out there that want to do just that. To me it would not be surprising if all of a sudden people can't use cars anymore, because someone has be slowly rolling out an attack and waiting to execute it for maximum effect.