Message from @Abe Lover

Discord ID: 648276166381273088


2019-11-24 21:34:32 UTC  

unbiased

2019-11-24 21:34:38 UTC  

As in “hi have a nice day “ or “ go F yourself”

2019-11-24 21:34:47 UTC  

still no

2019-11-24 21:35:05 UTC  

@everyone Ron Shekelson's accts have been shoad. SEE above video

2019-11-24 21:35:06 UTC  

its terms for if an ai humanitys goals includeed in its own goals or not

2019-11-24 21:35:07 UTC  

You can program something o appear friendly

2019-11-24 21:35:15 UTC  

That’s the artificial part

2019-11-24 21:35:19 UTC  

if an ai has not then it might kill us off as unneeded

2019-11-24 21:35:33 UTC  

an AI cant be described with those characteristics
its not a human
if its telling you to fuck yourself, likely it picked it up some from a dataset

2019-11-24 21:35:38 UTC  

if our goals are included it at least wont just get rid of us humans

2019-11-24 21:35:40 UTC  

Exactly

2019-11-24 21:35:52 UTC  

tahts why i explained it

2019-11-24 21:36:09 UTC  

the tearms just mean if the goal of the ai contains the goals of humanity like freedom etc

2019-11-24 21:36:26 UTC  

those seem like highly unspecific goals

2019-11-24 21:36:42 UTC  

yes they are cause the damn problem is what are the goals to set

2019-11-24 21:36:48 UTC  

tahts the whole damn problem

2019-11-24 21:37:03 UTC  

The more you know about AI the less you fear that it could take over the world

2019-11-24 21:37:14 UTC  

how do you make an ai care for humans and not let us end in a way we dont desire

2019-11-24 21:37:33 UTC  

again a very unspecific goal so good luck training it

2019-11-24 21:37:53 UTC  

i already said thats the whole problem

2019-11-24 21:38:02 UTC  

Ai can't be trained

2019-11-24 21:38:02 UTC  

and there lots of people working on this

2019-11-24 21:38:17 UTC  

They are only programmed to do what the human allows it to do

2019-11-24 21:38:25 UTC  

Movies aren't like ai

2019-11-24 21:38:28 UTC  

i told google alexa to destroy humanity, HELP

2019-11-24 21:38:29 UTC  

Machine learning can “train” it

2019-11-24 21:38:47 UTC  

today you train ais basically

2019-11-24 21:38:50 UTC  

It’s spooky because it seems real but it’s all artificial

2019-11-24 21:39:02 UTC  

ther eneuronal networks and at the end you got no idea why they do what they do

2019-11-24 21:39:19 UTC  

OK boomer

2019-11-24 21:39:26 UTC  

Yeah we program this neural networks and don’t really know how they work

2019-11-24 21:39:44 UTC  

no we let it learn from datasets

2019-11-24 21:39:53 UTC  

and then we dont know how it comes to its conclusions

2019-11-24 21:39:56 UTC  

https://cdn.discordapp.com/attachments/484514023698726912/648276642522857472/MVIMG_20191124_141716.jpg

2019-11-24 21:39:59 UTC  

Yeah

2019-11-24 21:40:22 UTC  

Didn't even give me a reason why I couldn't run away

2019-11-24 21:41:03 UTC  

the thing is hvaing an highly unspecific goal will make it harder for whoever is creating it to program it that way @negatic
we cant feed it something subjective like "take care of human"
and be done with it, unless you specifically define the actions and it goes on for a number of cycles repeating the task its not going to achieve anything

2019-11-24 21:41:25 UTC  

thats exactly what im saying satan

2019-11-24 21:41:49 UTC  

Deleting one line if code can f--k up the whole thing

2019-11-24 21:41:51 UTC  

thats what makes it very unlikely to create an AI like that

2019-11-24 21:41:56 UTC  

troll