AI Drone attacks human operator

[elguapo 4586]

Could we maybe not make Skynet happen irl?

 

[ztejas 43547]

It was a simulation, G.

[elguapo 4586]

2 minutes ago, ztejas said:

It was a simulation, G.

REALLY???

[Wally Fairway 5590]

12 minutes ago, ztejas said:

It was a simulation, G.

Shall  we play a game?

[crash_davis 23115]

The second part is even better. After telling it to NOT kill the operator, it destroys the communication tower so that the operator cannot tell it to NOT kill the operator.

 

[ztejas 43547]

23 minutes ago, elguapo said:

REALLY???

Well your title is extremely misleading. 

Not that it doing it in a simulation isn't alarming. 

Edited June 1, 2023 by ztejas

[BillyGoatHill 4312]

Yeah, this is going to turn out well.

[Willfully Horn 8057]

Folks who work in AI suck.

[Brisketexan 105446]

2 minutes ago, Willfully Horn said:

Folks who work in AI suck.

If we were AI drones, we would correctly see them as a threat to our survival and take them out.

Allegedly.

[Texaskimo 135]

Well the drone wasn't wrong in its threat assessment. 

[Upgrayedd 5821]

15 hours ago, ztejas said:

It was a simulation, G.

It was a thought experiment 

 

[BabaYaga 15328]

39 minutes ago, Upgrayedd said:

It was a thought experiment 

 

[hayden_horn 11365]

ai can't kill us unless we let it.  shit like this is how we die 

[hayden_horn 11365]

see? i can't even fucking post a gif from mobile and we are supposed to win this war

[BabaYaga 15328]

21 minutes ago, hayden_horn said:

see? i can't even fucking post a gif from mobile and we are supposed to win this war

[83Horn 1648]

 

[Helobious 9467]

Lulz at all the tech bros trying to downplay what happened. This shit is scary as fuck and I think global governments need to step in and put a stop to it

 

[AeroHorn 755]

Yeah, the A"I" is not going to attack the operator or the communications tower when it has been trained only on target data. What is scary is the automation and industrialization of killing without even a token moral agent (human soldier) involved in the decision-making. As it is, the "corporate-ness" of the military with greed and power-driven political directives combined with "follow without questions" trained soldiers eliminates human moral agency anyway. What AI will do is take this to its extreme and logical limit where the only moral agent left is the previous day's kill efficiency data. God help us.

[Dahobbs 17699]

40 minutes ago, Helobious said:

Lulz at all the tech bros trying to downplay what happened. This shit is scary as fuck and I think global governments need to step in and put a stop to it

What is it that you think happened? Your twitter quotes indicate that absolutely nothing happened. It was all just a hypothetical thought experiment, not an actual AI. 

[Wally Fairway 5590]

12 minutes ago, AeroHorn said:

Yeah, the A"I" is not going to attack the operator or the communications tower when it has been trained only on target data. What is scary is the automation and industrialization of killing without even a token moral agent (human soldier) involved in the decision-making. As it is, the "corporate-ness" of the military with greed and power-driven political directives combined with "follow without questions" trained soldiers eliminates human moral agency anyway. What AI will do is take this to its extreme and logical limit where the only moral agent left is the previous day's kill efficiency data. God help us.

sounds like these might have been Hammer drones

[hayden_horn 11365]

17 minutes ago, AeroHorn said:

Yeah, the A"I" is not going to attack the operator or the communications tower when it has been trained only on target data. What is scary is the automation and industrialization of killing without even a token moral agent (human soldier) involved in the decision-making. As it is, the "corporate-ness" of the military with greed and power-driven political directives combined with "follow without questions" trained soldiers eliminates human moral agency anyway. What AI will do is take this to its extreme and logical limit where the only moral agent left is the previous day's kill efficiency data. God help us.

so this is an amazing comment. somehow combined with your avatar, it's triggered something in my mind, but i'm not sure yet what.

your point that we are more worried about the thing going full

than the fact that it's capable of dispatching a human life without a real thought about it is monstrous.

[Upgrayedd 5821]

You guys realize that this was a thought experiment, right?

[miguelito 11314]

Half the posts in this thread were written by AI.

 

You decide which half.

[Fudge Nuggets 19227]

21 hours ago, crash_davis said:

The second part is even better. After telling it to NOT kill the operator, it destroys the communication tower so that the operator cannot tell it to NOT kill the operator.

 

That is all sorts of bad ass.

[Captainant 27833]

48 minutes ago, miguelito said:

Half the posts in this thread were written by AI.

 

You decide which half.

Chrispy hasn't posted THAT much in this thread yet

[Paco 1934]

1 hour ago, AeroHorn said:

As it is, the "corporate-ness" of the military with greed and power-driven political directives combined with "follow without questions" trained soldiers 

The fuck are you talking about? 

[WBT 20674]

We've got this

[Blotto 23584]

3 hours ago, Dahobbs said:

What is it that you think happened? Your twitter quotes indicate that absolutely nothing happened. It was all just a hypothetical thought experiment, not an actual AI. 

You would have a more productive conversation with a potato.

[AeroHorn 755]

3 hours ago, hayden_horn said:

so this is an amazing comment. somehow combined with your avatar, it's triggered something in my mind, but i'm not sure yet what.

your point that we are more worried about the thing going full

than the fact that it's capable of dispatching a human life without a real thought about it is monstrous.

I don't think it will go full autonomous enemy of humankind entity but it will be far more insidious, targeted, embedded, data-driven, inference+kill engine. Just one scenario: a tiny drone programmed to look for certain characteristics and autonomously complete its mission with only a probability percentage number and maybe a target signal to serve as extra training data.The drone can be launching a missile to take out a village or shooting a poisoned dart in the neck, and they could number in ones or millions.

The underlying danger is that the moral agency is further and further stratified to just a handful of people instead of at least hundreds of thousands of soldiers at a visible level now. So, a small senate committee can ask a secret three letter agency to do far more havoc than a war, and we would be insensitive to it. As it is, drones are killing weekly and we are largely uncaring. The dehumanization of humankind and being reduced to just a bio-chemical material is the tragic path we are on.

[AeroHorn 755]

2 hours ago, Paco said:

The fuck are you talking about? 

Killing is an industry and efficiency, throughput, cost basis, future growth, etc are the corporate metrics. AI will deliver across the board. Big thumbs up from the upper management.

[Sawbonz 27605]

3 hours ago, miguelito said:

Half the posts in this thread were written by AI.

 

You decide which half.

I can’t wait until we get to the point where AI can read the posts on surly for me and respond.  That way I can get some shit done.

Edited June 2, 2023 by Sawbonz
Can’t even do talk to text right at this point

[Brisketexan 105446]

1 minute ago, Sawbonz said:

I can’t wait until we get to the point where AI can read the posts on surly for me and respond.  That way I can get some shit done.

When AI starts responding with "rules" and "this thread is worthless without pics," we're either done for.....or we'll be fine, because the AI will be totally distracted looking at boobs.

[futureman 22554]

has this story been posted anywhere reputable?  is it even real?  or fake news.