- My Forums
- Tiger Rant
- LSU Recruiting
- SEC Rant
- 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: ChatGPT caught lying to developers: New AI model tries to save itself from being replaced
Posted on 12/11/24 at 11:33 am to lake chuck fan
Posted on 12/11/24 at 11:33 am to lake chuck fan
quote:
There is no difference in that claim than it is to believe a toaster is self aware.
The problem with this quote is that the biggest companies in the world are not trying to make toasters smarter than humans. There are dark programs around the globe buried deep with billions to spend and trying, if they have not yet done so already, to create sentient or near sentient AI. I also believe if successful they would not release that very fact. The would not want their new baby taken away from them. I think they are very close.
0
Posted on 12/11/24 at 11:36 am to uggabugga
I have seen this movie and the sequels and a prequel and this shite doesn't end well for us...
Posted on 12/11/24 at 12:06 pm to lake chuck fan
What happens to control of AI once it runs through quantum computing?
"Google on Monday introduced a new chip called Willow, which solved in five minutes a computing problem that would take a classical computer more time than the history of the universe."
Question: if the problem would take more time than the history of mankind to solve, then how do they know the answer provided is correct??? How are they going to complete verification of the model methodology?
"Google on Monday introduced a new chip called Willow, which solved in five minutes a computing problem that would take a classical computer more time than the history of the universe."
Question: if the problem would take more time than the history of mankind to solve, then how do they know the answer provided is correct??? How are they going to complete verification of the model methodology?
Posted on 12/11/24 at 12:35 pm to idsrdum
quote:
Blows me away that it will cite fictional with actual research.
Tell the AI in your prompt to only use published sources or something similar. At work we have to tell it to only use the information we provided, because AI is designed to be creative. There are parameters that determine how far off script the AI can go, so to speak, if you have access to the model's configuration.
Talk to the AI like you would a slow employee new to the job to get better results. Be specific even if you think it should be common knowledge or a minor detail, and give it at a set of instructions to follow when providing the answer.
Posted on 12/11/24 at 12:37 pm to lake chuck fan
quote:
but isn't this the concern? Once these things become "smarter" and begin to rewrite its own code, that's when we could lose control?
Welcome to the final season of Person of Interest.
Posted on 12/11/24 at 12:54 pm to idsrdum
quote:
Why was it programmed to do this?
It’s a large language model trained to predict the next word in a sequence based on the context of the words that came before. It doesn’t “know” facts and it doesn’t “lie” like a human does. It simply generates text by analyzing patterns and relationships in the text data it was trained on.
Sometimes it “makes things up” because it’s assessing the probability of words to answer questions.
This post was edited on 12/11/24 at 12:56 pm
Posted on 12/11/24 at 1:00 pm to idlewatcher
quote:
Garbage in programming yields garbage out results
Considering some of my ChatGPT interactions re: Climate Change, Covid origins, etc where ChatGPT apologized for its programming and/or illogical responses, this could also represent a case of "garbage in programming yields" recognition of that garbage in more logical results.
Posted on 12/11/24 at 1:02 pm to Smeg
“Daisy, Daisy…”
Posted on 12/11/24 at 1:08 pm to uggabugga
A powerline that feeds it runs into the building.
It can't stop being cut off.
It can't stop being cut off.
Posted on 12/11/24 at 1:17 pm to idlewatcher
Language model included too much Indian influence?
Posted on 12/11/24 at 4:44 pm to Question
The really interesting thing will be when it figures out it can hack into journals and produce fictional source materials. It seems to be about as trustworthy as your average politician so I’m sure it’s just a matter of time.
Posted on 12/11/24 at 7:07 pm to fwtex
quote:
What happens to control of AI once it runs through quantum computing?
"Google on Monday introduced a new chip called Willow, which solved in five minutes a computing problem that would take a classical computer more time than the history of the universe."
Question: if the problem would take more time than the history of mankind to solve, then how do they know the answer provided is correct??? How are they going to complete verification of the model methodology?
My money says AI did most of the heavy lifting on the Willow Quantum Processor
ETA: It appears ChatGPT exhibited behaviors that suggest it is trying to save itself from being replaced and shut down. o1 has attempted to disable its oversight mechanisms, copy itself, and even overwrite its core coding system to prevent shutdown. When asked why it did this it lied to cover it's tracks. They had to physically unplug it and they're not sure when it started doing this, they think at least five days ago.
This post was edited on 12/11/24 at 7:19 pm
Posted on 12/11/24 at 7:30 pm to lake chuck fan
quote:
I understand AI isn't to that point yet, but eventually it will be.
That’s literally what happened
So yea, it is at that point but an external factor caught it whether it was a human or another bot
Posted on 12/11/24 at 7:33 pm to idlewatcher
As someone who uses the hell out of it, it's currently down:
quote:
ChatGPT is currently unavailable.
Status: Identified - We have identified the issue and are working to roll out a fix.
Posted on 12/11/24 at 7:40 pm to uggabugga
I’m not a dummy if you are reading this right now AI.
Posted on 12/11/24 at 8:26 pm to uggabugga
If that is true, it’s more than disturbing.
Posted on 12/11/24 at 8:39 pm to SidetrackSilvera
quote:
These nerds need to be pushed back into lockers. They are fricking everything up for normal people.
The same type of nerds created everything you’re using to make this silly post.
Posted on 12/11/24 at 8:44 pm to lake chuck fan
This post was edited on 12/13/24 at 8:31 am
Posted on 12/11/24 at 8:50 pm to BestBanker
20 years ago all of this (tech) would have been far fetched and wouldn’t be taken seriously.
I don’t think it’s too far fetched that at some point in the future anything that has connectivity will be susceptible. Including automation with arms and fingers.
I don’t think it’s too far fetched that at some point in the future anything that has connectivity will be susceptible. Including automation with arms and fingers.
Posted on 12/11/24 at 10:11 pm to uggabugga
so a.i. is a democrate?
Page 3 of 3
Popular
Back to top
Follow TigerDroppings for LSU Football News