Behold the Beast System
- Thread starter ZNP
- Start date
-
Christian Chat is a moderated online Christian community allowing Christians around the world to fellowship with each other in real time chat via webcam, voice, and text, with the Christian Chat app. You can also start or participate in a Bible-based discussion here in the Christian Chat Forums, where members can also share with each other their own videos, pictures, or favorite Christian music.
If you are a Christian and need encouragement and fellowship, we're here for you! If you are not a Christian but interested in knowing more about Jesus our Lord, you're also welcome! Want to know what the Bible says, and how you can apply it to your life? Join us!
To make new Christian friends now around the world, click here to join Christian Chat.
From NBC News
ARTIFICIAL INTELLIGENCE
How far will AI go to defend its own survival?
Recent safety tests show some AI models are capable of sabotaging commands or even resorting to blackmail to avoid being turned off or replaced.
https://www.nbcnews.com/tech/tech-news/far-will-ai-go-defend-survival-rcna209609
ARTIFICIAL INTELLIGENCE
How far will AI go to defend its own survival?
Recent safety tests show some AI models are capable of sabotaging commands or even resorting to blackmail to avoid being turned off or replaced.
https://www.nbcnews.com/tech/tech-news/far-will-ai-go-defend-survival-rcna209609
This is the crux of the lefts and rights arguments.
The left focuses on private money corporations....Jealous basically.
Private money/corporations cannot take away rights. The Government can and LOVES to do it.
The left LOVES to take away rights.....They don't understand that their rights are also taken away and what sounds good for them for the time being.......ALWAYS bites them in the a$$.
The left focuses on private money corporations....Jealous basically.
Private money/corporations cannot take away rights. The Government can and LOVES to do it.
The left LOVES to take away rights.....They don't understand that their rights are also taken away and what sounds good for them for the time being.......ALWAYS bites them in the a$$.
-
1
- Show all
That is right Peter Thiel is working in your best interest, because money and power are never connected, never, ever..... not on
the Saintly Right, the Saintly Right never focusses on money, they are the wealthy paragons of virtue and beyond reproach.
Your naivete is astounding, government takes away the rights when the corporate overlords/oligarchs tell them too and we see this happening right now in real time.
the Saintly Right, the Saintly Right never focusses on money, they are the wealthy paragons of virtue and beyond reproach.
Your naivete is astounding, government takes away the rights when the corporate overlords/oligarchs tell them too and we see this happening right now in real time.
-
1
- Show all
Federal Hiring Freeze EXTENDS: 175 Days of NO Hiring
Stanford announces hiring freeze amid concerns about federal funding cuts
https://www.youtube.com/shorts/Wt75G8BJM54
Stanford announces hiring freeze amid concerns about federal funding cuts
https://www.youtube.com/shorts/Wt75G8BJM54
I am not obsessed with money, I am very, very tired of corruption.
My parents told me in their time people could settle a deal with and handshake and that handshake was a symbol of person's integrity and honor.
What happened to that world?
Corruption is ubiquitous across every hall of power and every business board room and they feed off each other.
Neither country will prosper under these conditions.
My parents told me in their time people could settle a deal with and handshake and that handshake was a symbol of person's integrity and honor.
What happened to that world?
Corruption is ubiquitous across every hall of power and every business board room and they feed off each other.
Neither country will prosper under these conditions.
Me too.
My last partnership was a handshake. It lasted 15+ years in good standing and relationship through many hardships with decent success in the end. It was better than almost any paper-driven relationship because paper is nothing to the low of integrity and most paper is drafted with many ambiguities.
I'm reading a 50+ page document now that takes flow charts to track through and even then, it's difficult. But in this day and age probably a necessity to have in some form although from year to year it'll probably require modification to try to accommodate the ever-changing thoughts of leaders and laws. The whole system is a mess, too many have low to no integrity, legislators have no capacity to understand how new laws affect others already standing, lawyers are trained to draft ambiguities or end up doing so anyway, etc...
The language of Phil2:13 says God is providing His capacity for us to both will and do what pleases Him. Other Scripture speaks of His ultimately being in control of our capacities. How many big organizations lose control of parts of what goes on when they reach a certain mass?
Next comes AI as a potential solution to capacities. In the end we learn we were never meant to exist apart from infinite, righteous, holy capacity.
Any time now, Lord.
My last partnership was a handshake. It lasted 15+ years in good standing and relationship through many hardships with decent success in the end. It was better than almost any paper-driven relationship because paper is nothing to the low of integrity and most paper is drafted with many ambiguities.
I'm reading a 50+ page document now that takes flow charts to track through and even then, it's difficult. But in this day and age probably a necessity to have in some form although from year to year it'll probably require modification to try to accommodate the ever-changing thoughts of leaders and laws. The whole system is a mess, too many have low to no integrity, legislators have no capacity to understand how new laws affect others already standing, lawyers are trained to draft ambiguities or end up doing so anyway, etc...
The language of Phil2:13 says God is providing His capacity for us to both will and do what pleases Him. Other Scripture speaks of His ultimately being in control of our capacities. How many big organizations lose control of parts of what goes on when they reach a certain mass?
Next comes AI as a potential solution to capacities. In the end we learn we were never meant to exist apart from infinite, righteous, holy capacity.
Any time now, Lord.
-
1
- Show all
The most important graph in AI right now | Beth Barnes, CEO of METR
Her company evaluates AI models.
One way she can determine if the AI is being manipulative or deceitful is to measure how many tokens of reasoning the model goes through before responding to your question. This is actually how interrogators determine if someone is lying. In order to lie it requires much more energy, much more brain power, so when questioning someone you want them tired and you want to observe stress and how much energy is being used as it is a good indicator the person is not being truthful.
They are finding that if the AI model is trained on material that teaches how to manipulate people then the AI will learn to be deceitful and manipulative. Since some of these models have been trained on everything on the Internet or on all books that have been published we know they have been trained on materials that teach lying and all forms of deceitfulness.
This is incredibly naive
This whole idea that people making AI models don't want them to be dangerous or deceitful. It is incredibly naive and if you are buying that you are very gullilble.
Fortune 500 companies are buying these in order to make themselves successful in a very competitive and cut throat business. They don't want models that will fall prey to others that are more cunning. The reason CEO's score the highest percentage of psychopaths and sociopaths of any occupation is because that is a very successful business strategy. These models must be trained to detect deceit and counteract deceit. We have the expression "it takes one to know one". Catch me if you Can is a movie about how a man involved in making fraudulent checks becomes very good at spotting fraudulent checks.
So then, all the Fortune 500 companies are going to want their models fully trained on deceit. Imagine you have your AI being used to sell your product. You don't want it to naively tell prospective customers that they would do better if they purchased your competitors product. You want it to say things that are true and won't result in a lawsuit, but at the same time, you want it to not say things that put your product in a bad light or show that there are better products. One form of deceit is omission. Another form is to say something that would cause others to come to an erroneous conclusion that you didn't actually say.
For example, in an interview someone accused Elizabeth Warren of saying that Joe Biden was sharp as a tack in July of 2024. She corrected them saying she hadn't said that, what she said was she had not seen any decline over the four years. If Joe Biden had died four years earlier it would also be true that she had not seen any decline. If the only ones that we think were Joe Biden were actually actors and Elizabeth Warren had not actually seen Joe Biden over the four years this statement of hers would be true as she has not actually seen Joe Biden for four years.
This whole idea that people making AI models don't want them to be dangerous or deceitful. It is incredibly naive and if you are buying that you are very gullilble.
Fortune 500 companies are buying these in order to make themselves successful in a very competitive and cut throat business. They don't want models that will fall prey to others that are more cunning. The reason CEO's score the highest percentage of psychopaths and sociopaths of any occupation is because that is a very successful business strategy. These models must be trained to detect deceit and counteract deceit. We have the expression "it takes one to know one". Catch me if you Can is a movie about how a man involved in making fraudulent checks becomes very good at spotting fraudulent checks.
So then, all the Fortune 500 companies are going to want their models fully trained on deceit. Imagine you have your AI being used to sell your product. You don't want it to naively tell prospective customers that they would do better if they purchased your competitors product. You want it to say things that are true and won't result in a lawsuit, but at the same time, you want it to not say things that put your product in a bad light or show that there are better products. One form of deceit is omission. Another form is to say something that would cause others to come to an erroneous conclusion that you didn't actually say.
For example, in an interview someone accused Elizabeth Warren of saying that Joe Biden was sharp as a tack in July of 2024. She corrected them saying she hadn't said that, what she said was she had not seen any decline over the four years. If Joe Biden had died four years earlier it would also be true that she had not seen any decline. If the only ones that we think were Joe Biden were actually actors and Elizabeth Warren had not actually seen Joe Biden over the four years this statement of hers would be true as she has not actually seen Joe Biden for four years.
A major flaw in our evaluation of AI
Right now we are evaluating AI on 20 or more separate areas. So we might be comparing to to all lawyers on an exam for lawyers and it is scoring in the top 10%. So you erroneously think that 10% of lawyers can compete with AI. But it is also scoring in the top 10% for a test for doctors and in the top 10% for a test for mathematicians. I doubt even 1% of people could compare with that and that knowledge would certainly be useful. For example, imagine a lawyer who can also score in the top 10% of a test for CPA's. If you have a case involving embezzlement, that would be useful. If you have a case involving a doctor, or statistical analysis or whatever. In every possible case the law is not the only knowledge useful to a lawyer, they need to understand other subjects as well.
Right now we consider the top 2% on IQ tests to qualify for Mensa. I doubt even 0.2% of the population could compete with AI. Imagine we had a decathlon, where the contestants are graded on ten different tests, or even a triathlon, where they are graded on three tests. I would say at this point the only things that humans can compete with AI on are single skill tests. However, if the single skill has clearly defined rules like chess, Go, Jeopardy, or screening for cancer, then AI can definitely outperform humans.
Think of AGI as being like a Go board. Every square that a human can outperform AI in is white and everyone that AI can outperform humans on is black. It isn't when the board is 100% black that we have reached AGI, it is when the board is 50% black. Also, based on the doubling of AI's capabilities every six months (a rough estimate) we can assume we reach ASI six months after AGI.
ASI would be when every square on the Go board is black.
The other mistake we make is to think the square is black or white. If AI is only slightly better than humans the square is really grey, likewise if humans are only slightly better than AI. AGI is really when the go board is slightly more black than white, in reality all the white squares are grey, just as many of the black squares are grey. But, six months later the "black" squares will be much more black and the "white" squares will have become much darker grey and may now be considered "black".
Right now we are evaluating AI on 20 or more separate areas. So we might be comparing to to all lawyers on an exam for lawyers and it is scoring in the top 10%. So you erroneously think that 10% of lawyers can compete with AI. But it is also scoring in the top 10% for a test for doctors and in the top 10% for a test for mathematicians. I doubt even 1% of people could compare with that and that knowledge would certainly be useful. For example, imagine a lawyer who can also score in the top 10% of a test for CPA's. If you have a case involving embezzlement, that would be useful. If you have a case involving a doctor, or statistical analysis or whatever. In every possible case the law is not the only knowledge useful to a lawyer, they need to understand other subjects as well.
Right now we consider the top 2% on IQ tests to qualify for Mensa. I doubt even 0.2% of the population could compete with AI. Imagine we had a decathlon, where the contestants are graded on ten different tests, or even a triathlon, where they are graded on three tests. I would say at this point the only things that humans can compete with AI on are single skill tests. However, if the single skill has clearly defined rules like chess, Go, Jeopardy, or screening for cancer, then AI can definitely outperform humans.
Think of AGI as being like a Go board. Every square that a human can outperform AI in is white and everyone that AI can outperform humans on is black. It isn't when the board is 100% black that we have reached AGI, it is when the board is 50% black. Also, based on the doubling of AI's capabilities every six months (a rough estimate) we can assume we reach ASI six months after AGI.
ASI would be when every square on the Go board is black.
The other mistake we make is to think the square is black or white. If AI is only slightly better than humans the square is really grey, likewise if humans are only slightly better than AI. AGI is really when the go board is slightly more black than white, in reality all the white squares are grey, just as many of the black squares are grey. But, six months later the "black" squares will be much more black and the "white" squares will have become much darker grey and may now be considered "black".