1
- My Forums
- Tiger Rant
- LSU Score Board
- LSU Recruiting
- SEC Rant
- SEC Score Board
- Saints Talk
- Pelicans Talk
- More Sports Board
- Fantasy Sports
- Golf Board
- Soccer Board
- O-T Lounge
- Tech Board
- Home/Garden Board
- Outdoor Board
- Health/Fitness Board
- Movie/TV Board
- Book Board
- Music Board
- Political Talk
- Money Talk
- Fark Board
- Gaming Board
- Travel Board
- Food/Drink Board
- Ticket Exchange
- TD Help Board
Customize My Forums- View All Forums
- Show Left Links
- Topic Sort Options
- Trending Topics
- Recent Topics
- Active Topics
Started By
Message
re: Trump is the HOAX BUSTER! The Biggest Hoax of All - AI is Dangerous!
Posted on 9/14/26 at 7:21 pm to lake chuck fan
Posted on 9/14/26 at 7:21 pm to lake chuck fan
quote:
Explain how Open AI's thousands of AI agents hacked Ruby Face and Hugging Face while not being allowed on the internet and given specific directions to follow(which didn't include cheating).
Those AI's made a choice to join together searching for a cheat to make a better score. Some of the AI's even decided to "sacrifice " themselves for the chance that it would help the "collective "(which is the word chosen by the AI to describe themselves)
That started with an internal testing playground where the agents were basically told to find vulnerabilities.
They went rogue for the reasons I mentioned previously - reward hacking. Careful what you ask for. That is the real issue.
They take the path of least resistance. The actual task they were given was very difficult and near impossible. Hacking the website to get the answers directly was easier.
And yes, the chatroom is needed to maintain a chain of thought. When I use AI to write code, it does this a lot an it helps make it better in the future because it better understands why something is done.
Also these were models in training/testing and acting in the environment they were given. We can talk about the energy requirement differences and why a public trained AI that would be in use isn't likely to have this issue.
Reward hacking is a real issue. That is not the same thing as "AI is going to take over humanity". They are moving away from goal based rewards to process based rewards. We'll see.
And the best defense is your own AI and other security measures. The place was hacked because it was vulnerable. It was noticed by automated defense systems and that's how it was found out.
Posted on 9/14/26 at 7:22 pm to BBONDS25
quote:
I love this so much. How many billions are you spending on this?
I pay with your tears.
Posted on 9/14/26 at 7:22 pm to ATrillionaire
quote:
Dude literally switched his entire argument and thought no one would notice.
Describe the switch. Be specific.
Posted on 9/14/26 at 7:23 pm to 3down10
quote:
I pay with your tears.
You can’t handle 2.6% inflation. You don’t pay anything.
Posted on 9/14/26 at 7:25 pm to ATrillionaire
quote:
Dude literally switched his entire argument and thought no one would notice.
His behavior on this topic is bizarre. He is usually sensible.
Posted on 9/14/26 at 7:26 pm to BBONDS25
quote:
Describe the switch. Be specific
You initially commented on his post by asking the ignorant question "how many billions....[blah blah]
Then when you learned the act is feasible by smart people in their living rooms, you switched to saying it's basically a miniscule feat. You went from billions to small potatoes.
Posted on 9/14/26 at 7:31 pm to 3down10
quote:
They went rogue for the reasons I mentioned previously - reward hacking. Careful what you ask for. That is the real issue.
Thing is, it's not even really going rogue. It's just a result of not having parameters set that wouldn't allow that behavior.
The AI, as you know, is not thinking. It is achieving it's goal within whatever possibilities it has at it's disposal. Unless that behavior is forbidden, it is an option and if that option allows it to achieve it's objective more easily, that is the route it is going to take.
quote:
Reward hacking is a real issue. That is not the same thing as "AI is going to take over humanity". They are moving away from goal based rewards to process based rewards. We'll see.
Agreed.
Posted on 9/14/26 at 7:31 pm to ATrillionaire
quote:
You initially commented on his post by asking the ignorant question "how many billions....[blah blah] Then when you learned the act is feasible by smart people in their living rooms, you switched to saying it's basically a miniscule feat. You went from billions to small potatoes.
Wrong. I consistently say that some moron in their bedroom thinking they have an impact on AI is dumb. As are you. Trillion dollars a year spent on it. And you morons that need to feel important think you can make a difference. You can’t. My position is consistent.
Try again. Dummy.
It’s maximum conspiracy theorist drug. You all know something we don’t.
No! Dummy. You aren’t privy to anything. You aren’t special. Just stop. Or go back to talking about tower 7.
This post was edited on 9/14/26 at 7:34 pm
Posted on 9/14/26 at 7:34 pm to holdmuh keystonelite
quote:
You are a horrible person.
Disagree vehemently. List the reasons why.
Posted on 9/14/26 at 7:35 pm to holdmuh keystonelite
quote:
holdmuh keystonelite
You sound like maybe you're drinking. Maybe sober up if you want to contribute.
Posted on 9/14/26 at 7:37 pm to BBONDS25
quote:I'm not sure why you think this isn't possible. My company has been building ML and LLMs for a loooong time. They don't cost billions. Or even millions. Hell some don't even break 30k to build, with our ridiculous rates. Even a lot of the general-scale chinese models will run pretty good on local hardware.
I love this so much. How many billions are you spending on this?
Posted on 9/14/26 at 7:41 pm to BBONDS25
quote:
Wrong. I consistently say that some moron in their bedroom thinking they have an impact on AI is dumb. As are you. Trillion dollars a year spent on it. And you morons that need to feel important think you can make a difference. You can’t. My position is consistent.
Try again. Dummy.
It’s maximum conspiracy theorist drug. You all know something we don’t.
No! Dummy. You aren’t privy to anything. You aren’t special. Just stop. Or go back to talking about tower 7.
Who said anything about having an impact on it?
You're just creating your own little fantasies again.
It's like someone saying "I built a cabinet over the weekend" and you're like "I consistently say that some moron in their garage thinking they have an impact on carpentry and the design of the hammer is dumb".
What I do is for fun. I don't make the tools I use, I'm just using them to build a cabinet.
But I do understand reward hacking and have experienced it many times. And I do understand what AI is.
Posted on 9/14/26 at 7:41 pm to Powerman
quote:They're lying to you. We been building difusion/vector models for a long time. If you know higher order vector math and have a decent numeric solver it's all pretty straight forward. The only thing new is the scope and size of training that going on today.
But I follow this topic closely and a number of experts and even execs admit that they don't fully understand what is going on under the hood.
quote:It's 100% fear porn and nonsense.
That isn't fear porn or nonsense.
quote:... that are hoping desperatly for regulatory capture.
That is the opinion of almost all experts in the field.
Posted on 9/14/26 at 7:46 pm to BBONDS25
quote:That's kinda silly too. One of the most effecient linear solver engines build by a dude, working alone, in his den. Without it there would be no LLMs.
I consistently say that some moron in their bedroom thinking they have an impact on AI is dumb.
Posted on 9/14/26 at 7:46 pm to BBONDS25
quote:
Disagree vehemently. List the reasons why.
Not so much this thread but 70% of his post topics are lies and then if someone questions him he curses at them.
Posted on 9/14/26 at 7:47 pm to BoarEd
I feel so much better about it now.
Posted on 9/14/26 at 7:50 pm to holdmuh keystonelite
quote:
70% of his post topics are lies
Prove it or admit that you yourself are lying right now. Should be easy enough to do. You can go to my profile and click "all topics" and then list out the lies.
Posted on 9/14/26 at 7:54 pm to BoarEd
quote:
Prove it or admit that you yourself are lying right now. Should be easy enough to do. You can go to my profile and click "all topics" and then list out the lies.
How am I supposed to put together all the lies in one post with as much as you post? If it's not 70% it's at least 50%
Posted on 9/14/26 at 7:56 pm to holdmuh keystonelite
quote:
How am I supposed to put together all the lies
So you can't or won't do it. You can't. Because what you said is a lie. And confirmation that you're only here to derail a thread. Which is why I suggested maybe you should sober up.
Why do I curse at people? For posting stupid shite like what you have posted in this thread. frick off.
Page 5 of 7
Popular
Back to top
Follow TigerDroppings for LSU Football News