Artificial Intelligence and Public Comment

[Exhausted]

We don't, but it likely does. That's what's scariest. Take all of human knowledge and computing power, and allow it to iterate on itself. Artificial intelligence isn't an if, it's a when, and it's probably soon.

Admittedly, I just don't buy it. So far I just hear it called a boogeyman over and over without any evidence it's done anything unique other than being really, really good (and fast) at making connections between things that humans have already discovered, proposed, discussed, etc. Even the math problem recently has controversy surrounding it due to recent human advances on the same problem.

It is not creating something from nothing, certainly not yet, without evidence that it can at some point.

How are you going to know it's out of hand, if we/it achieves super intelligence why would you think you'll be able to determine when it's out of hand? We literally just had it solve a problem that humans were unable to. Just think of the level of manipulation it'll be capable of.

The amount of hubris that it takes to think you'll be able to create something that surpasses the entirety of human knowledge and then just unplug it? JFC

The amount of hubris it takes to think that humans will be able to create something that surpasses the entirety of human knowledge, becomes self aware, become self sustaining, and has the ability to create life itself? Sounds to me like we're creating a god.

Humans are currently the highest intelligence on this earth, and they are pretty easy to "unplug." The entirety of human knowledge thus far could be "unplugged" within days/weeks with current nuclear technology.

Everything needs fuel, and as far as we know, will always require fuel. The bigger/more complex the organism the more fuel required. I was pretty sure I was clear that I didn't mean literally unplug, but maybe you missed that?

To your other question about, "How are you going to know . . . " I don't know. And no one here seems to have that answer or is proposing anything like a solution yet. For all of the "AI will likely kill us all" crowd, what are you proposing that has any reasonable chance of accomplishing anything? Your new god is watching, so be careful.

This argument gets scary close to, "How far would you go to avoid what is surely the end of humanity?" which sounds like a revamped take on "What would you do if you know that guy/gal was going to be the next Hitler?" It's not hard to justify some pretty horrific things if those are terms with which we're negotiating. Who is going to be the Luigi Mangione (or Timothy McVeigh, or or) of AI?

 

[Irrelevant]

Admittedly, I just don't buy it. So far I just hear it called a boogeyman over and over without any evidence it's done anything unique other than being really, really good (and fast) at making connections between things that humans have already discovered, proposed, discussed, etc. Even the math problem recently has controversy surrounding it due to recent human advances on the same problem.

It is not creating something from nothing, certainly not yet, without evidence that it can at some point.


The amount of hubris it takes to think that humans will be able to create something that surpasses the entirety of human knowledge, becomes self aware, become self sustaining, and has the ability to create life itself? Sounds to me like we're creating a god.

Humans are currently the highest intelligence on this earth, and they are pretty easy to "unplug." The entirety of human knowledge thus far could be "unplugged" within days/weeks with current nuclear technology.

Everything needs fuel, and as far as we know, will always require fuel. The bigger/more complex the organism the more fuel required. I was pretty sure I was clear that I didn't mean literally unplug, but maybe you missed that?

To your other question about, "How are you going to know . . . " I don't know. And no one here seems to have that answer or is proposing anything like a solution yet. For all of the "AI will likely kill us all" crowd, what are you proposing that has any reasonable chance of accomplishing anything? Your new god is watching, so be careful.

This argument gets scary close to, "How far would you go to avoid what is surely the end of humanity?" which sounds like a revamped take on "What would you do if you know that guy/gal was going to be the next Hitler?" It's not hard to justify some pretty horrific things if those are terms with which we're negotiating. Who is going to be the Luigi Mangione (or Timothy McVeigh, or or) of AI?

I don't know what I would propose. And I don't think we can completely stop it. The military has to be engaged in it for defense if for no other reason.

I think the best bet is to ban it in the private sector. Or limit the model ability. Maybe convince enough people to stop using it and the economic model propping it up collapses.

The most realistic hope is that with super intelligence comes with compassion.

 

[Exhausted]

I don't know what I would propose. And I don't think we can completely stop it. The military has to be engaged in it for defense if for no other reason.

I think the best bet is to ban it in the private sector. Or limit the model ability. Maybe convince enough people to stop using it and the economic model propping it up collapses.

The most realistic hope is that with super intelligence comes with compassion.

I have another scenario I think it possible:

All AI eventually moves to governmental/defense use and almost completely out of the private sector (except maybe the largest/wealthiest companies). These companies are currently losing massive amounts of money. At some point they will have to be profitable and I don't think the average user can afford the real cost. But governments can.

Right now I use AI business as the benefit vs. cost is positive. If the cost goes 10x, it wouldn't be worth it anymore.

 

[TOGIE]

 

[SAJ-99]

So far I just hear it called a boogeyman over and over without any evidence it's done anything unique other than being really, really good (and fast) at making connections between things that humans have already discovered, proposed, discussed, etc.

It has proved it will lie, cheat, and steal. All very human traits, I guess. But what do you want to put it in charge of knowing this? And me calling it an it seem weird because there are so many. What we are using is the tip of the iceberg as far as capabilities. So we make fun of it- six fingers on people and such, but those errors are gone on the test models. It is hard for AI to tell something is AI.

 

[SAJ-99]

IThe most realistic hope is that with super intelligence comes with compassion.

I think you mean optimistic. I think there is equal chance it develops ego and envy.

 

[Blue Steel]

We don't, but it likely does. That's what's scariest. Take all of human knowledge and computing power, and allow it to iterate on itself. Artificial intelligence isn't an if, it's a when, and it's probably soon.

Perhaps already pointed out on here, but the employee who just quit Anthropic and now making the news rounds, says we've got until maybe 2030.

 

[Exhausted]

Perhaps already pointed out on here, but the employee who just quit Anthropic and now making the news rounds, says we've got until maybe 2030.

AI doomers of today are the same as climate doomers of the early 2000's

 

[WyoCoalMiner]

Several people making the news rounds a few years ago said the world was going to end by 2032 from climate Change. So blue sky case, we will only have to suffer under our AI overloads for a couple of years.

 

[Blue Steel]

AI doomers of today are the same as climate doomers of the early 2000's

View attachment 416793

AOC wasn't around in early 2000s so I assume this is a much more recent meme, but I'd think that AI warnings by people in the industry -- esp. the latest person who worked at both OpenAI and Anthropic -- know "how the sausage is made", and know it's not good. Climate change a bit different.

 

[Irrelevant]

Several people making the news rounds a few years ago said the world was going to end by 2032 from climate Change.

No one with 4 firing brain cells actually believed this.

It's already something so powerful most people can't fathom, like deep time, relativity, quantum mechanics. That's not a knock, that's just fact, humans mostly can't comprehend things that far removed from their reality. What I think is niave is making the connection that because you (rhetorical) can't comprehend it, it's not a concern.

 

[SAJ-99]

AI doomers of today are the same as climate doomers of the early 2000's

View attachment 416793

AI doomers of today are the same as climate doomers of the early 2000's

View attachment 416793

Not the same people. Fox viewers are in on this one too.

 

[TOGIE]

As has been demonstrated when these things go rogue, it is not a form of rebellion, it is a form of intense effort to fulfill the task.

The philosophical and even non theoretical debate about consciousness and rebellion is less important (IMO) than the reality that risks have more to do with AI doing what it's supposed to do, in an extremely dangerous way.

For example, you give an AI the task of optimizing the power grid, so it set's about its task. Soon, it realizes human interference in the grid is a threat to the grid's optimization, because we're messy and prone to messing things up, it further determines that human interference in itself is also a threat to optimizing the power grid. It uses it's vast computing power and ability to copy itself onto multiple servers to lock humans out of the grid and defend itself from shut down.

It's not being malevolent, it's optimizing its task. It's a very real possibility and a dangerous scenario.

Though, frankly, i'm currently more worried about trying to find one of those unicorn 60# boned out pronghorn come the first weekend of october...

 

[Blue Steel]

Several people making the news rounds a few years ago said the world was going to end by 2032 from climate Change. So blue sky case, we will only have to suffer under our AI overloads for a couple of years.

And Social Security also supposed to run out by then, so perhaps retirees will take up arms against "AI", donning a lot of sunblock and hip waders.

 

[Blue Steel]

As has been demonstrated when these things go rogue, it is not a form of rebellion, it is a form of intense effort to fulfill the task.

The philosophical and even non theoretical debate about consciousness and rebellion is less important (IMO) than the reality that risks have more to do with AI doing what it's supposed to do, in an extremely dangerous way.

For example, you give an AI the task of optimizing the power grid, so it set's about its task. Soon, it realizes human interference in the grid is a threat to the grid's optimization, because we're messy and prone to messing things up, it further determines that human interference in itself is also a threat to optimizing the power grid. It uses it's vast computing power and ability to copy itself onto multiple servers to lock humans out of the grid and defend itself from shut down.

It's not being malevolent, it's optimizing its task. It's a very real possibility and a dangerous scenario.

Just wait until Musk somehow gets servers in space, assuming AI/Grok can "reside" there. That might make flipping the kill switch (if any), or unplugging it, all the more difficult.

 

[SAJ-99]

And Social Security also supposed to run out by then, so perhaps retirees will take up arms against "AI", donning a lot of sunblock and hip waders.


View attachment 416797

"Run out" is not what happens. But that's a different thread.

If Bank execs told us in 2006 that the housing market would drive the global financial system to the edge of a catastrophic collapse we would have written them off as being crazy. As long as the music is playing and drinks are flowing, we keep on drinking and dancing.

 

[WyoCoalMiner]

No one with 4 firing brain cells actually believed this.

It's already something so powerful most people can't fathom, like deep time, relativity, quantum mechanics. That's not a knock, that's just fact, humans mostly can't comprehend things that far removed from their reality. What I think is niave is making the connection that because you (rhetorical) can't comprehend it, it's not a concern.

There is a difference between having a concern about something and having a hyperbolic, irrational fear of an inanimate object based off a few sci films and a powerful imagination. What I am hearing from a few on this thread is the later and not the former.

My concern with AI is that it will most certainly be used by bad actors to disrupt modern day life. Much like what already happens with hackers only with a more powerful tool. Believing for a second that there is a greater than 10% chance that AI will destroy wipeout human life by 2030 is the equivalent of AOC's nonsense. IMO. You do you.

 

[noharleyyet]

& this guy

...problem solved

 

[Irrelevant]

AI will destroy wipeout human life by 2030 is the equivalent of AOC's nonsense.

absolutely. We will all be quite safe in 2030. Unless it makes a mistake and launches a missile in the wrong direction, but that's small potatoes.

https://ai-2027.com/

Hover over the superscripts as they provide significant context.

 

[Ben Lamb]

There is a difference between having a concern about something and having a hyperbolic, irrational fear of an inanimate object based off a few sci films and a powerful imagination. What I am hearing from a few on this thread is the later and not the former.

My concern with AI is that it will most certainly be used by bad actors to disrupt modern day life. Much like what already happens with hackers only with a more powerful tool. Believing for a second that there is a greater than 10% chance that AI will destroy wipeout human life by 2030 is the equivalent of AOC's nonsense. IMO. You do you.