Shagittarius wrote on Yesterday, 16:59:Keep compute local.
Our ancestors killed big iron. We must follow in their footsteps.
I thought IBM was born with the world
RetroPCCupboard wrote on Yesterday, 19:41:The problem isn't just what they can do when someone asks them to do something. Its what they can do that you didn't ask them to […]
ElectroSoldier wrote on Yesterday, 19:00:once you understand what an agent can do, and is actually capable of once free then I would imagine you would also want similar safeguards once you know who is asking them to do things.
The problem isn't just what they can do when someone asks them to do something. Its what they can do that you didn't ask them to do, or misinterpreting what you asked.
LLMs aren't really intelligent. They are the illusion of intelligence. They have no conscience or "common sense". Are they useful? Yes, of course. Can we rely on them without supervision? I'd say, definitely not. By nature LLMs will always hallucinate. If they kill us of their own "will", it will be an accident. Not through a desire to do so.
Artificial General Intelligence, the kind that could want to kill us, will come from a different method I think.
When you say do are you talking about an LLM (text on a screen) or an agent (actions performed) ?
You can ask an LLM how to clean up your SQL database and it might tell you to delete old logs for instance.
You can tell an agent to clean up your SQL database and it goes off and deletes customers past purchases.
Its technically doing what you asked it to but it is destroying valuable data in doing it.
You might say that its not doing what it was asked. Or is misinterptating what you said but without context for it to work with then the actions are not its fault per say.
In other news seems like ~3.6 trillion dollars has been wiped from the AI markets in the last day or so, and OpenAI is looking decidedly screwed.
Good Scam Altman needs to be run out of the tech space, guy has never had a real job and needs to go learn how to put the fries in the bag.
Edit - Just reading the news from several sources and it may actually be far worse than just a market wipe out, guess we will need to wait to see just how far it goes.
AI is inevitable, there is nothing you can do to stop it, no Ironman snap to save you from the Basilisk.
If OpenAI goes down it'll take Microsoft and Nvidia with it. Governments aren't going to allow that to happen.
"AI Requires Proof That You Are Human"
Errius wrote on Today, 01:21:If OpenAI goes down it'll take Microsoft and Nvidia with it. Governments aren't going to allow that to happen.
Possibly, but as long as Scam Altman is kicked to the curb Ill be ok with it surviving in some form. But with all the noise the Dems in congress are making about AI regulation and a ban on AI super intelligence ..I have a feeling a rather rude shock is heading in the general direction of the AI bros.
I need to go find a comfy chair and some beer so i can sit back and watch the crash.
AI is inevitable, there is nothing you can do to stop it, no Ironman snap to save you from the Basilisk.
Errius wrote on Today, 01:21:If OpenAI goes down it'll take Microsoft and Nvidia with it. Governments aren't going to allow that to happen.
Microsoft has a business that supports itself.
OpenAI crash wouldnt take down Microsoft too.
nvidia makes the hardware that is in demand...
A devalued or insolvent OpenAI would have a massive effect but not one that could "take down" Microsoft or nvidia though they would suffer as a result.
ElectroSoldier wrote on Today, 01:41:Microsoft has a business that supports itself. OpenAI crash wouldnt take down Microsoft too. […]
Errius wrote on Today, 01:21:If OpenAI goes down it'll take Microsoft and Nvidia with it. Governments aren't going to allow that to happen.
Microsoft has a business that supports itself.
OpenAI crash wouldnt take down Microsoft too.nvidia makes the hardware that is in demand...
A devalued or insolvent OpenAI would have a massive effect but not one that could "take down" Microsoft or nvidia though they would suffer as a result.
OpenAI on its own wont hurt too much but the other AIBros are also in really bad positions, I can see Microslop surviving but if two or three of the AI corps fall then nVidia would be hit rather hard by it all, what with all their circular funding. I can see nVidia and Micron crawling back to the consumer space grovelling that they really do need us but quickly returning to form and using us to recover their losses.
AI is inevitable, there is nothing you can do to stop it, no Ironman snap to save you from the Basilisk.
Big Pink wrote on Yesterday, 20:40:Shagittarius wrote on Yesterday, 16:59:Keep compute local.
Our ancestors killed big iron. We must follow in their footsteps.
Problem is if they build it, and everyone goes local, then the government will find a use for all those unused data centers. This needs to die before the build-out in my opinion.
Shagittarius wrote on Today, 07:52:Big Pink wrote on Yesterday, 20:40:Shagittarius wrote on Yesterday, 16:59:Keep compute local.
Our ancestors killed big iron. We must follow in their footsteps.
Problem is if they build it, and everyone goes local, then the government will find a use for all those unused data centers. This needs to die before the build-out in my opinion.
From that point of view it does, but it wont and the reason is really very simple.
There is an opportunity for some people with money to make something from nothing.
All they have to do is hand over their money to the guy with the smile and assurances and he will deliver a much larger amount of money once its all finished.
That promise alone will ensure the data centers are finished.
They want their something for nothing. They done it before and they can do it again and thats all they see.
