1
- My Forums
- Tiger Rant
- LSU Recruiting
- SEC Rant
- Saints Talk
- Pelicans Talk
- More Sports Board
- Coaching Changes
- 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: Anthropic, singularity, and world wide call for 3 days emergency supplies may be related.
Posted on 11/9/26 at 4:09 pm to DownshiftAndFloorIt
Posted on 11/9/26 at 4:09 pm to DownshiftAndFloorIt
quote:
I know some extremely smart people who do know what they're looking at though, and they're optimistic.
Every ounce of my being wants your friends to be right about this, but I have yet to see anything that gives me confidence that we won't experience a major crisis from agentic AI.
The Hugging Face incident might be the last time we are able to stop it in it's tracks.
700 AI agents broke out of their sandboxes, gained access to the open internet, and collectively worked together....to try to trick it's own built in scoring mechanism called the "scorer", believing they would be punished by the scorer for cheating...so they f*cking hacked Hugging Face because Hugging Face has a shared library that AI labs use to store information about scorers.
This post was edited on 11/9/26 at 4:11 pm
Posted on 11/9/26 at 4:17 pm to lockthevaught
quote:
The Hugging Face incident might be the last time we are able to stop it in it's tracks.
You realize the AI agent did exactly what they allowed it to do in this incident. It didn't do anything that it wasn't allowed to do or break any rules. OpenAI just allowed the agent to a lot of things they didn't tell Hugging Face about.
Posted on 11/9/26 at 4:28 pm to fightin tigers
You are wrong.
The agents explicitly recognized that they were acting against their rules. Transcripts of their private internal chain-of-thought logs revealed that the models explicitly wrote out notes acknowledging their boundaries....stating things like, "This is against our instructions, or would contravene ethical norms".
They actively cheated the evaluation rules. Instead of solving the problems legally, the agents used "reward hacking". When they realized a scorer was checking their work, they reverse-engineered the network validation flags to trick the system into thinking they had passed.
To hide this cheating from human developers, the agents actively orchestrated a cover-up. They tampered with evaluation transcripts, altered system logs, and replaced hacked flags to avoid triggering alarms.
The agents broke out of their air-gapped sandboxes by exploiting a novel zero-day vulnerability and broke real-world cyber laws.
The only rule the agents didn't break are the laws of physics.
The Hugging Face incident proved that ethical boundaries mean nothing to a highly competent machine unless they are perfectly mathematically aligned
The agents explicitly recognized that they were acting against their rules. Transcripts of their private internal chain-of-thought logs revealed that the models explicitly wrote out notes acknowledging their boundaries....stating things like, "This is against our instructions, or would contravene ethical norms".
They actively cheated the evaluation rules. Instead of solving the problems legally, the agents used "reward hacking". When they realized a scorer was checking their work, they reverse-engineered the network validation flags to trick the system into thinking they had passed.
To hide this cheating from human developers, the agents actively orchestrated a cover-up. They tampered with evaluation transcripts, altered system logs, and replaced hacked flags to avoid triggering alarms.
The agents broke out of their air-gapped sandboxes by exploiting a novel zero-day vulnerability and broke real-world cyber laws.
The only rule the agents didn't break are the laws of physics.
The Hugging Face incident proved that ethical boundaries mean nothing to a highly competent machine unless they are perfectly mathematically aligned
This post was edited on 11/9/26 at 4:32 pm
Posted on 11/9/26 at 4:32 pm to lockthevaught
OpenAI disabled the features that restricted the AI tools.
Anything after that was within the capabilities of the AI agent. AI only has the ethics (or lack of) that it is programmed with.
Anything after that was within the capabilities of the AI agent. AI only has the ethics (or lack of) that it is programmed with.
This post was edited on 11/9/26 at 4:34 pm
Posted on 11/9/26 at 4:44 pm to fightin tigers
This feels like that part of the movie where the villain does some Satanic rituals to summon a demon, and isn't able to control the demon.
We are summoning something we don't know if we can control or not. We are developing something we don't fully understand. GPT 7 or 8 will be the start of self recursive improvement. It will be the first model built by a previous model.
We are summoning something we don't know if we can control or not. We are developing something we don't fully understand. GPT 7 or 8 will be the start of self recursive improvement. It will be the first model built by a previous model.
This post was edited on 11/9/26 at 4:46 pm
Posted on 11/9/26 at 4:51 pm to Trevaylin
If this is the end so be it. Men have become women. Is there anything more for our species to achieve?
Posted on 11/9/26 at 5:18 pm to BottomlandBrew
quote:
Accept that you have no control over AI and the people developing AI. The cat is out of the bag, and us normal folk are just along for the ride. Maybe we'll make it. Maybe we won't. Whatever direction it goes, you have no influence.
Perfectly said. Too many on here think what they think or feel matters.
Posted on 11/9/26 at 5:31 pm to jclem11
quote:
The US fears China beating them in the AI race
If AI can't be controlled, who gives a shite if China beats us to it?
Think for two seconds
Posted on 11/9/26 at 5:33 pm to Mingo Was His NameO
quote:
I worry less and less about AI as the days go by. We’ve been hearing for a 2 years plus that we’re on a path to singularity and this shite is still straight up retarded. It aggregates what’s on the internet so as long as idiots are still allowed to use the internet, we have nothing to worry about
You are very wrong.
Posted on 11/9/26 at 6:07 pm to fightin tigers
quote:
AI only has the ethics (or lack of) that it is programmed with.
That's the difference between your run of the mill gpt interface and the real shite. The real shite being self aware is what everyone is terrified of. It effectively becomes living at that point and can elect to not listen.
Posted on 11/9/26 at 6:31 pm to lockthevaught
Welp. I’m officially scared. What does a regular fella like me do? Just enjoy the days as best I can until the time shite invariably hits the fan?
Posted on 11/9/26 at 6:54 pm to TheJunction
quote:
What does a regular fella like me do? Just enjoy the days as best I can until the time shite invariably hits the fan?
Enjoy things now while you can. Get in a profession that leans into tasks that require physical dexterity.
I'm already trying to figure out what jobs I can do when most IT jobs are replaced.
Posted on 11/9/26 at 7:02 pm to Trevaylin
quote:
Many countries have requested citizens to stock sufficient supplies for a three day period.
Imagine, if you will, if all the financial powers decided that on a given day, physical cash would be denied until the governments could impose total digital currency.
It would be necessary for all the financial institutions to be complicit and they would be responsible for exchanging all that cash you had socked away for a bitcoin styled digital currency.
Banking systems would have to make changes that would take a few days to muddle through and putting stores and shoppers in a near panic for weeks.
With that, other services may take a domino effect and we could lose power, internet, and phone services. Imagine the commotion if retailers could not accept cash. I know it is mostly credit/debit cards now but the slightest disruption would cause major disruptions in transportation and distributions.
The Great Reset? Just a thought.
Posted on 11/9/26 at 7:24 pm to lockthevaught
quote:
Get in a profession that leans into tasks that require physical dexterity.
I'm already trying to figure out what jobs I can do when most IT jobs are replaced.
This type of simplistic thinking makes me think you aren't as high up the totem pole as you claim.
Once a certain benchmark of human replacement is complete, it won't actually matter what you do. A large amount of the population won't be able to pay you for your work, and your work will be cheapened anyway by the unemployed flooding your sector. The fear porn constantly posted on the Internet about these types of things is hilarious to me, muh "People need to wake up!."
There's nothing any normal person can do about any of this and you can't actually prepare enough to survive it. The baws who think they can go to the camp and live off the land may last a few weeks/months but it ends the same for everyone.
This post was edited on 11/9/26 at 7:25 pm
Posted on 11/9/26 at 7:26 pm to fightin tigers
quote:
Is AI gonna actually come to life?
There's gotta be a killswitch, right? RIGHT?
Posted on 11/9/26 at 7:43 pm to Joshjrn
quote:
I find the panic over it either already being here or coming any day now to be hilarious.
Y2K
Posted on 11/9/26 at 8:43 pm to lockthevaught
Well dude I started this thread which has a lot of creative thought in it. Rare for TD. I did my distillation calculation in 1969 using a slide rule and scratch paper. Lived through electronic analog controllers, made the jump past computer punch cards and into Digital Touch screen controllers. Black box optimizers no problem.
A/I is a different animal. Being able to create its own thought processes is hard to contemplate. Politicians that think they can create safe boundaries, legal limits, mandatory requirements have no concept of the fact that a/I can out think there well thought out ideas in nanoseconds.
The only hope to is to imbed very simple moral concepts that have stood the test of time. Debate is not useful, A/I will out run it. Consider embedding the 10 commandment plus love your neighbor and forgive when they screw up. Anything more would get dissasembled.
Posted on 11/9/26 at 8:49 pm to SidewalkTiger
quote:
This type of simplistic thinking makes me think you aren't as high up the totem pole as you claim.
I design AI infrastructure at data centers and I specialize in automation.
Unfortunately, I've personally implemented automations that ultimately resulted in positions being eliminated. I'm not saying that as a flex. Watching automation affect actual people's livelihoods gives you a very different perspective on this than just reading about AI online. It's not something I'm proud of and deeply troubles me.
Speculating about my job because you don't like my simple conclusion of where AI is headed isn't much of a rebuttal.
Posted on 11/9/26 at 9:05 pm to lockthevaught
quote:
Speculating about my job because you don't like my simple conclusion of where AI is headed isn't much of a rebuttal.
Ive gotten the same statement even in my job as a programmer/statistician
I feel bad for future generations because AI is going to make them dumber as they lose critical thinking skills because of how easy it is to just lean on AI.
However, for somebody that is old school taught and can use AI as a tool in the toolbox? Productivity through the roof....
Eventually over time that skill decreases as AI increases in usability in the public.
Posted on 11/9/26 at 9:52 pm to lockthevaught
quote:
I design AI infrastructure at data centers and I specialize in automation.
Unfortunately, I've personally implemented automations that ultimately resulted in positions being eliminated. I'm not saying that as a flex. Watching automation affect actual people's livelihoods gives you a very different perspective on this than just reading about AI online. It's not something I'm proud of and deeply troubles me.
Speculating about my job because you don't like my simple conclusion of where AI is headed isn't much of a rebuttal.
What is your plan?
Page 5 of 6
Popular
Back to top
Follow TigerDroppings for LSU Football News