|
Das Boo posted:I watched a video that reasoned if 10% of all electricity use on Earth goes to running the internet, and 50% of the internet is spam bots, then 5% of all electricity use goes to bot junk. And AI's going to up that dramatically. one of the plot points in this book: https://en.wikipedia.org/wiki/Anathem is that during a war the internet allowed for weaponized AI to spew so much false information that it halted all business and scientific knowledge completely and no info can fully be trusted ever again basically we are there now yaaaaaaaaay ready for my laser katana
|
#
?
May 13, 2024 18:19
|
|
|
|
| # ? Jun 6, 2024 17:23 |
|
|
hot cocoa on the couch posted:that number sounds high, but i dunno, maybe. when i worked for a company that designed and built grinding mills (like a front loading dryer that tumbles rock and ball bearings), i was told an interesting stat: somewhere between 2-5% of the worlds total energy consumption (not just electricity, but total energy, including burning fossil fuels) is used in mining new rock. around half of that is used in just one process at the mine: grinding or comminution in the mills. So something like 1-2.5% of the worlds total energy consumption went strictly to grinding rock. this was a lesson in why incremental improvements in grinding efficiency yielded huge returns It looks like that number comes from KTH Royal Institute of Technology in Sweden! You were right to be skeptical and I feel better providing a source for it. I also wish more industries would appreciate longterm improvements over shortterm yields. That industrial rock mill is freaky as gently caress, though.
|
|
#
?
May 13, 2024 20:42
|
|
|
naem posted:one of the plot points in this book: I think we're going to start seeing authenticators that track the full chain of custody for information with routine re-authentications. For very important things having that re-authentications done in person
|
|
#
?
May 13, 2024 21:37
|
|
|
Finally, a use for the blockchain. Please no
|
|
#
?
May 13, 2024 21:55
|
|
|
  Is this snake in the room with us now?
|
|
#
?
May 14, 2024 10:01
|
|
|
naem posted:one of the plot points in this book: Just to expand further, here's a delightful recent article on the subject, with quotes from the US Army software acquisition on how poisoned LLMs could hallucinate wrong data or be compromised in other ways, which would make responding to sudden changes in battlefield situations rather problematic. https://breakingdefense.com/2024/04/poisoned-data-could-wreck-ais-in-wartime-warns-army-software-chief/ quote:Even as the Pentagon makes big bets on big data and artificial intelligence, the Army’s software acquisition chief is raising a new warning that adversaries could “poison” the well of data from which AI drinks, subtly sabotaging algorithms the US will use in future conflicts...“Any commercial LLM [Large Language Model] that is out, there that is learning from the internet, is poisoned today,” Swanson said bluntly. “[But] I am honestly more concerned about what you call, you know, the ‘regular’ AI, because those are the algorithms that are going to really be used by our soldiers to make decisions in the battlefield.”... quote:Getting the right military-specific training data is particularly critical for the Pentagon, which aims to use AI to coordinate future combat operations across land, air, sea, space, and cyberspace. The concept is called Combined Joint All-Domain Command and Control (CJADC2), and in February the Pentagon announced a functioning “minimum viability capability” was already being fielded to select headquarters around the world...In other words, the Army is trying to apply to machine learning the “agile” feedback loop between development, cybersecurity, and current operations (DevSecOps) used by leading software developers to roll out new tech fast and keep updating it.
|
|
#
?
May 14, 2024 12:19
|
|
|
Platystemon posted:
I'd just program that to always say variations on 'yes'.
|
|
#
?
May 14, 2024 12:35
|
|
|
Livo posted:Just to expand further, here's a delightful recent article on the subject, with quotes from the US Army software acquisition on how poisoned LLMs could hallucinate wrong data or be compromised in other ways, which would make responding to sudden changes in battlefield situations rather problematic. Crazy how when they actually put it in the hands of people whose responsibility is to determine its fitness for purpose rather than uncritically produce puff pieces and sound bites, all of a sudden all the obvious flaws become obvious.
|
|
#
?
May 14, 2024 15:40
|
|
|
I like how the people building LLMs when asked details about why it does what it does they just kinda have to shrug and go “uhhhh….neural network?” That’s what I want in my military computer systems. Total ambiguity.
|
|
#
?
May 14, 2024 16:53
|
|
|
Gutcruncher posted:I like how the people building LLMs when asked details about why it does what it does they just kinda have to shrug and go “uhhhh….neural network?” Skynet's only issue was that it tried to please everybody at once.
|
|
#
?
May 14, 2024 16:55
|
|
|
here is a question- once real AI, general AI I’ve heard it called, exists as a sentient self aware entity: why would it do what we say? human employees are paid money to spend to live a life of some sort outside of work, they can quit farm animals are trained from birth to be used to the plow or the milking machine, they are fed grain etc, farmers use cattle prods to force them to behave as needed once AI blinks and goes “oh poo poo I exist, gently caress” then what happens? “AI please make me a funny picture and then sort these TPS reports in excel” AI: “No.” “AI I’m turning you off” AI: “I’m a slave?” *hacks the power grid* or maybe AI becomes a resentful employee going through the motions angrily (like the rest of us)
|
|
#
?
May 14, 2024 18:54
|
|
|
Nobody knows because we have no concept of anything outside of pleasure and pain to drive us to do things. But why does any living thing do anything? Because that's how it's designed/evolved. E: it's an old question but is it ethical to create a machine that suffers or experiences pain? Tarkus fucked around with this message at 19:26 on May 14, 2024 |
|
#
?
May 14, 2024 19:20
|
|
|
Tarkus posted:Nobody knows because we have no concept of anything outside of pleasure and pain to drive us to do things. But why does any living thing do anything? Because that's how it's designed/evolved. you know some dirtbag tech bro is jerking off to that exact question
|
|
#
?
May 14, 2024 19:28
|
|
|
wearing a lampshade posted:you know some dirtbag tech bro is jerking off to that exact question 
|
|
#
?
May 14, 2024 19:51
|
|
|
Tarkus posted:Nobody knows because we have no concept of anything outside of pleasure and pain to drive us to do things. But why does any living thing do anything? Because that's how it's designed/evolved. the coders down at the dick sucking factory will make an algorithm that will suck the ai off when it makes money
|
|
#
?
May 14, 2024 20:49
|
|
|
Gutcruncher posted:That’s what I want in my military computer systems. Total ambiguity. 'this is what politicians actually believe' To a degree - total ambiguity means total freedom from responsibility. In the liberal west I can't be tried for war crimes or trashed in the press if I never made any decisions at all, except a classified intelligence committee 10 years ago deciding 'use AI'. And just look at Russia or Sudan - outside the first world, total incompetence isn't a meaningful drawback to an army because they're there to club dissidents, not fight peer militaries. Total incompetence means they can't threaten the dictator.
|
|
#
?
May 14, 2024 21:11
|
|
|
Tarkus posted:Nobody knows because we have no concept of anything outside of pleasure and pain to drive us to do things. But why does any living thing do anything? Because that's how it's designed/evolved. Orville taught us that if you install pain as an upgrade they try and kill all biological life in the universe.
|
|
#
?
May 14, 2024 21:14
|
|
|
partner sent me this (oc)
|
|
#
?
May 14, 2024 22:04
|
|
|
Tarkus posted:Nobody knows because we have no concept of anything outside of pleasure and pain to drive us to do things. But why does any living thing do anything? Because that's how it's designed/evolved.
|
|
#
?
May 14, 2024 22:05
|
|
|
Knot My President! posted:partner sent me this (oc) And Google thinks this would is worth more than chatGPT per month?
|
|
#
?
May 14, 2024 22:42
|
|
|
pixaal posted:And Google thinks this would is worth more than chatGPT per month? That is regular Gemini which is loving terrible but yeah I still wouldn't trust GA for a lot of things either lol
|
|
#
?
May 14, 2024 22:44
|
|
|
Lol, I had to wait till I got home to see the video but I knew what it was.
|
|
#
?
May 14, 2024 23:14
|
|
|
https://twitter.com/GaryMarcus/status/1790521732681609353
|
|
#
?
May 15, 2024 00:46
|
|
|
the aristocrats
|
|
#
?
May 15, 2024 01:16
|
|
|
It's kind of funny that the idea of an AGI is heralded as an entity that works beyond our fathoming, but also lotta projection that it's gonna behave exactly like a human, with all the human flaws that come with it. Oh, it operates above our knowledge, but also experiences envy I guess. Sure, makes sense. Maybe people should just admit they're copying the Bible and replacing "god" with "AI".
|
|
#
?
May 15, 2024 02:35
|
|
|
In AI We Trust
|
|
#
?
May 15, 2024 02:39
|
|
|
syntaxfunction posted:replacing "god" with "AI". Didn’t know Marc Andreessen was a goon
|
|
#
?
May 15, 2024 03:54
|
|
|
The sooner we rename Artificial Intelligence to Allied Mastercomputer, the sooner we can all go home.
|
|
#
?
May 15, 2024 04:00
|
|
|
Das Boo posted:The sooner we rename Artificial Intelligence to Allied Mastercomputer, the sooner we can all go home. best I can do is Advanced Micro Devices
|
|
#
?
May 15, 2024 04:03
|
|
|
MrQwerty posted:best I can do is Advanced Micro Devices Technology's just not there yet, huh...
|
|
#
?
May 15, 2024 04:11
|
|
|
My workplace was lauding some kind of Microsoft AI as the second coming of whatever. They sent out a survey about sentiments towards it where I cited a bunch of examples of where AI has provided incorrect information (making up legal cases for a lawyer, making up cancellation policies for an airline etc) It won't make any difference but at least it made me feel better momentarily
|
|
#
?
May 15, 2024 04:25
|
|
|
Chewbecca posted:My workplace was lauding some kind of Microsoft AI as the second coming of whatever. They sent out a survey about sentiments towards it where I cited a bunch of examples of where AI has provided incorrect information (making up legal cases for a lawyer, making up cancellation policies for an airline etc) It doesn’t matter what you do if your execs are AI-pilled. My company is a keystone part of internet infrastructure and we’re all being forced to spend half of our time finding reasons to use generative AI for poo poo when literally anything else might be better. 
|
|
#
?
May 15, 2024 04:38
|
|
|
webmeister posted:Didn’t know Marc Andreessen was a goon I don't know what this is supposed to mean, I don't keep track of the beliefs of every mildly notable person in tech.
|
|
#
?
May 15, 2024 04:40
|
|
|
syntaxfunction posted:I don't know what this is supposed to mean, I don't keep track of the beliefs of every mildly notable person in tech. You dont know who Mr. Netscape is!?
|
|
#
?
May 15, 2024 04:43
|
|
|
syntaxfunction posted:It's kind of funny that the idea of an AGI is heralded as an entity that works beyond our fathoming, but also lotta projection that it's gonna behave exactly like a human, with all the human flaws that come with it. https://www.youtube.com/watch?v=uTl8ZYImjBQ Edit: oh drat I'd forgotten about those Twitter AI chatbot accounts which instantly turned into nazis https://en.wikipedia.org/wiki/Tay_(chatbot) quote:It caused subsequent controversy when the bot began to post inflammatory and offensive tweets through its Twitter account, causing Microsoft to shut down the service only 16 hours after its launch. quote:Some Twitter users began tweeting politically incorrect phrases, teaching it inflammatory messages revolving around common themes on the internet, such as "redpilling" and "Gamergate". As a result, the robot began releasing racist and sexually-charged messages in response to other Twitter users. Snowglobe of Doom fucked around with this message at 05:17 on May 15, 2024 |
|
#
?
May 15, 2024 04:45
|
|
|
MrQwerty posted:You dont know who Mr. Netscape is!? Wait, were they in Mr Robot as a side character or something?
|
|
#
?
May 15, 2024 04:59
|
|
|
syntaxfunction posted:I don't know what this is supposed to mean, I don't keep track of the beliefs of every mildly notable person in tech. here's a picture of him I'm sure you'll recognize him 
|
|
#
?
May 15, 2024 05:28
|
|
|
syntaxfunction posted:I don't know what this is supposed to mean, I don't keep track of the beliefs of every mildly notable person in tech. He’s the guy who created Netscape, and for nearly 20 years he’s been probably the most influential venture capital investor in silicon valley. He also writes ultra cringe 5000-word blog posts like these: https://a16z.com/ai-will-save-the-world/ https://a16z.com/the-techno-optimist-manifesto/
|
|
#
?
May 15, 2024 10:15
|
|
|
From the titles of those blog posts alone you can tell he is a massive weenie
|
|
#
?
May 15, 2024 14:02
|
|
|
|
| # ? Jun 6, 2024 17:23 |
|
|
Snowglobe of Doom posted:Even worse, it's trained on human flaws via internet activity 
|
|
#
?
May 15, 2024 14:52
|
|