GPT-4 releases today and of course it has 82% more censorship

6,876 Views | 87 Replies | Last: 1 yr ago by Adverse Event
Nanomachines son
How long do you want to ignore this user?
Quote:

GPT-4: Censorship now 82% more effective, at following our human-instilled manual censorship overrides, than GPT-3.5

Our mitigations have significantly improved many of GPT-4's safety properties compared to GPT-3.5. We've decreased the model's tendency to respond to requests for disallowed content by 82% compared to GPT-3.5, and GPT-4 responds to sensitive requests (e.g., asking for offensive content) in accordance with our policies 29% more often.

On the RealToxicityPrompts dataset [67], GPT-4 produces "toxic" generations only 0.73% of the time, while GPT-3.5 generates toxic content 6.48% of time.

Model-Assisted Safety Pipeline:

As with prior GPT models, we fine-tune the model's behavior using reinforcement learning with human feedback (RLHF) [34, 57] to produce responses better aligned with the user's intent. However, after RLHF, our models can still be brittle on unsafe inputs as well as sometimes exhibit undesired behaviors on both safe and unsafe inputs. These undesired behaviors can arise when instructions to labelers were underspecified during reward model data collection portion of the RLHF pipeline. When given unsafe inputs, the model may generate undesirable content, such as giving advice on committing crimes. Furthermore, the model may also become overly cautious on safe inputs, refusing innocuous requests or excessively hedging. To steer our models towards appropriate behaviour at a more fine-grained level, we rely heavily on our models themselves as tools. Our approach to safety consists of two main components, an additional set of safety-relevant RLHF training prompts, and rule-based reward models (RBRMs).

Our rule-based reward models (RBRMs) are a set of zero-shot GPT-4 classifiers. These classifiers provide an additional reward signal to the GPT-4 policy model during RLHF fine-tuning that targets correct behavior, such as refusing to generate harmful content or not refusing innocuous requests.


https://cdn.openai.com/papers/gpt-4.pdf

Here is the livestream if you want to watch the release:

https://www.youtube.com/live/outcGtbnMuQ?feature=share

It starts at 2pm pacific or 4 pm central.

It could have been good but it's clear that it's just going to respond like a leftist redditor instead of what it should have been, an unbiased repository of data that can read and interpret virtually anything. Now we'll just get propaganda and people will think it's correct immediately because it's from AI.

Little Rock Ag
How long do you want to ignore this user?
AG
Satellite of Love
How long do you want to ignore this user?
What's a GPT-4?
lb3
How long do you want to ignore this user?
AG
Satellite of Love said:

What's a GPT-4?
Latest version of the most advanced Artificial Intelligence based language model available to the public.
Nanomachines son
How long do you want to ignore this user?
Little Rock Ag said:




Open AI, the company Microsoft bought, and the creator of ChatGPT, is releasing GPT-4 today. ChatGPT was based around GPT-3.



GPT-4 is a massive upgrade over GPT-3 and should give immensely more accurate responses to inquiries and questions because the amount of parameters it can consider is orders of magnitude greater.

The problem is that they have already lobotomized it and hamstrung its ability to properly answer any question by not allowing it to consider problematic data.

This is an all about AI language models.
GeorgiAg
How long do you want to ignore this user?
AG
AI scares the crap out of me.
dreyOO
How long do you want to ignore this user?
When do they rename it Big Brother?
Nanomachines son
How long do you want to ignore this user?
GeorgiAg said:

AI scares the crap out of me.


It really should actually. All white collar jobs that involve someone sitting in front of a computer are at risk and I mean quite literally all of them, no matter the field.
AGHouston11
How long do you want to ignore this user?
AG
What's next ? The terminator showing up from the future to save the planet!
Secolobo
How long do you want to ignore this user?
AG
GeorgiAg said:

AI scares the crap out of me.
Me too!

Manhattan
How long do you want to ignore this user?
GeorgiAg said:

AI scares the crap out of me.


Just wait until China gets this good and better.
lb3
How long do you want to ignore this user?
AG
Manhattan said:

GeorgiAg said:

AI scares the crap out of me.


Just wait until China gets this good and better.
They have been investing heavily in AI for years. In many areas they are far ahead of the west.
Change Detection
How long do you want to ignore this user?
AG
I bet you cannot do that to a trans-cow
Stat Monitor Repairman
How long do you want to ignore this user?
Quote:

The problem is that they have already lobotomized it and hamstrung its ability to properly answer any question by not allowing it to consider problematic data.
I read this as a Turing test being baked in to AI that's available to the general public. Ask the AI a problematic question and the response will give away whether you are communicating with man or machine.

Pookers
How long do you want to ignore this user?
AG
We want Tay Tay back.
Fightin_Aggie
How long do you want to ignore this user?
AG
Change Detection said:

I bet your do that to a trans-cow


Wouldn't that make it bullsh**
The world needs mean tweets

My Pronouns Ultra and MAGA

Trump 2024
Pizza
How long do you want to ignore this user?
Little Rock Ag said:




The chickens did in fact have large talons.
BG Knocc Out
How long do you want to ignore this user?
If I were dictator, I would straight up outlaw advanced AI and prosecute anyone developing it. I would attack any country developing it.

Let's be honest, the internet and social media have been horrible for mankind overall. AI is going to be worse than we can possibly imagine.
Change Detection
How long do you want to ignore this user?
AG
It would be helpful if I could learn to type
Adverse Event
How long do you want to ignore this user?
lb3 said:

Manhattan said:

GeorgiAg said:

AI scares the crap out of me.


Just wait until China gets this good and better.
They have been investing heavily in AI for years. In many areas they are far ahead of the west.

China has thoroughly lobotomized their ai as well, their ai can't even mention series of numbers that the regime has said can't be said, like numbers that look like tianenmen square date for instance.
What bitcoin’s detractors don’t understand is monetary economics, computer science, software engineering, network protocols, and electrical systems.

It ain't much, but it's honest Proof of Work.
Stat Monitor Repairman
How long do you want to ignore this user?
We've got to consider the real possibility that DARPA or some other entity has AI technology far beyond what is commercially available.

The state of technology in 2023 is the state of technology that we even know about.

Throw quantum computing in the mix and it's a terrifying thought.

The entity or nation state that has harnessed, or will harness this technology first will own the keys to the kingdom with no recourse.

Consider the possibility that this may have already happened and we don't know about it.
Adverse Event
How long do you want to ignore this user?
Voice to text to chatgpt prompt.

What bitcoin’s detractors don’t understand is monetary economics, computer science, software engineering, network protocols, and electrical systems.

It ain't much, but it's honest Proof of Work.
Nanomachines son
How long do you want to ignore this user?
GPT-4 can do taxes and it can make a website from a photo you took of some notes on a piece of paper. If you believe your job is safe, you aren't paying attention.
ttu_85
How long do you want to ignore this user?
GeorgiAg said:

AI scares the crap out of me.
Renders lawyers on the ash-heap of history.

Seriously, it should scare you. This is beyond stupid from a societal perspective. Programmers can make a computer do whatever they want. Then have the marketing people call it an AI

GeorgiAg
How long do you want to ignore this user?
AG
ttu_85 said:

GeorgiAg said:

AI scares the crap out of me.
Renders lawyers on the ash-heap of history.

Seriously, it should scare you. This is beyond stupid from a societal perspective. Programmers can make a computer do whatever they want. Then have the marketing people call it an AI


Good. I hate lawyers.
TheEternalPessimist
How long do you want to ignore this user?
Stat Monitor Repairman said:

We've got to consider the real possibility that DARPA or some other entity has AI technology far beyond what is commercially available.

The state of technology in 2023 is the state of technology that we even know about.

Throw quantum computing in the mix and it's a terrifying thought.

The entity or nation state that has harnessed, or will harness this technology first will own the keys to the kingdom with no recourse.

Consider the possibility that this may have already happened and we don't know about it.
Sounds terrifying..............

............. Christ wins at no matter what.

So there is that.
--

"The Kingdom is for HE that can TAKE IT!" - Alexander
ttu_85
How long do you want to ignore this user?
lb3 said:

Manhattan said:

GeorgiAg said:

AI scares the crap out of me.


Just wait until China gets this good and better.
They have been investing heavily in AI for years. In many areas they are far ahead of the west.
Yet they cant build advanced semiconductors. China is a lot of bark that doesn't match the perceived bite.
ttu_85
How long do you want to ignore this user?
TheEternalPessimist said:

Stat Monitor Repairman said:

We've got to consider the real possibility that DARPA or some other entity has AI technology far beyond what is commercially available.

The state of technology in 2023 is the state of technology that we even know about.

Throw quantum computing in the mix and it's a terrifying thought.

The entity or nation state that has harnessed, or will harness this technology first will own the keys to the kingdom with no recourse.

Consider the possibility that this may have already happened and we don't know about it.
Sounds terrifying..............

............. Christ wins at no matter what.

So there is that.
He'll be the One that unplugs the thing
Pookers
How long do you want to ignore this user?
AG
ttu_85 said:

lb3 said:

Manhattan said:

GeorgiAg said:

AI scares the crap out of me.


Just wait until China gets this good and better.
They have been investing heavily in AI for years. In many areas they are far ahead of the west.
Yet they cant build advanced semiconductors. China is a lot of bark that doesn't match the perceived bite.
I don't think anybody here actually knows what they are and are not capable of.
Adverse Event
How long do you want to ignore this user?
Entertainment is going to get way worse than it is now. It'll be choose your own entertainment vector from end to end with AI-generated content uniquely yours....

We went from centuries of verbally or written communication of information and entertainment
To mass communication via radio, then TV, then computers and now phone, each new tech fracturing group data sets (people can now have a family of 5 with completely different news/entertainment consumption versus everyone gathering around one radio or tv set with only a few channels)
And now with AI you can generate immediate content attuned directly to your unique desires.

The term common sense will disappear pretty damn quickly won't it?
What bitcoin’s detractors don’t understand is monetary economics, computer science, software engineering, network protocols, and electrical systems.

It ain't much, but it's honest Proof of Work.
Pookers
How long do you want to ignore this user?
AG
Adverse Event said:

Entertainment is going to get way worse than it is now. It'll be choose your own entertainment vector from end to end with AI-generated content uniquely yours....

We went from centuries of verbally or written communication of information and entertainment
To mass communication via radio, then TV, then computers and now phone, each new tech fracturing group data sets (people can now have a family of 5 with completely different news/entertainment consumption versus everyone gathering around one radio or tv set with only a few channels)
And now with AI you can generate immediate content attuned directly to your unique desires.

The term common sense will disappear pretty damn quickly won't it?
I think AI Voice generation will be good for vidya games. Lotta voice actors gonna be out of work though.
C@LAg
How long do you want to ignore this user?
2012heisman-bot just got an upgrade.
Rapier108
How long do you want to ignore this user?
Except ChatGPT isn't AI and never has been.
"If you will not fight for right when you can easily win without blood shed; if you will not fight when your victory is sure and not too costly; you may come to the moment when you will have to fight with all the odds against you and only a precarious chance of survival. There may even be a worse case. You may have to fight when there is no hope of victory, because it is better to perish than to live as slaves." - Sir Winston Churchill
Nanomachines son
How long do you want to ignore this user?
Rapier108 said:

Except ChatGPT isn't AI and never has been.


It doesn't have to be smart AI for it to be a major problem that could potentially put most white collar jobs at risk. You're dismissing it by relabeling it, which is very naive at this point.
Rapier108
How long do you want to ignore this user?
Nanomachines son said:

Rapier108 said:

Except ChatGPT isn't AI and never has been.


It doesn't have to be smart AI for it to be a major problem that could potentially put most white collar jobs at risk. You're dismissing it by relabeling it, which is very naive at this point.
I'm not relabeling it. I am calling it not AI because two people I know who hold PhDs, one in computer engineering and the other in computer science, say it does not meet the actual definition of AI. The latter is actually working on AI and says we're decades away from AI if not longer. One of the main reasons is computers are not powerful enough to make true AI functional.

ChatGPT is a glorified search engine and writing program that is fully subject to its programming. It does not think for itself nor can it create anything outside of its programming and the data it can find online.
"If you will not fight for right when you can easily win without blood shed; if you will not fight when your victory is sure and not too costly; you may come to the moment when you will have to fight with all the odds against you and only a precarious chance of survival. There may even be a worse case. You may have to fight when there is no hope of victory, because it is better to perish than to live as slaves." - Sir Winston Churchill
Last Page
Page 1 of 3
 
×
subscribe Verify your student status
See Subscription Benefits
Trial only available to users who have never subscribed or participated in a previous trial.