Posted on 12/17/2024 9:40:19 PM PST by BenLurkin
World models take inspiration from the mental models of the world that humans develop naturally. Our brains take the abstract representations from our senses and form them into more concrete understanding of the world around us, producing what we called “models” long before AI adopted the phrase. The predictions our brains make based on these models influence how we perceive the world.
...
While the concept has been around for decades, world models have gained popularity recently in part because of their promising applications in the field of generative video.
...
While a generative model trained on years of video might accurately predict that a basketball bounces, it doesn’t actually have any idea why — just like language models don’t really understand the concepts behind words and phrases. But a world model with even a basic grasp of why the basketball bounces like it does will be better at showing it do that thing.
To enable this kind of insight, world models are trained on a range of data, including photos, audio, videos, and text, with the intent of creating internal representations of how the world works, and the ability to reason about the consequences of actions.
...
Meta chief AI scientist Yann LeCun says the models could...could help achieve a desired goal through reasoning.
...
They could also spawn more capable robots.
Robots today are limited in what they can do because they don’t have an awareness of the world around them (or their own bodies). World models could give them that awareness, Mashrabov said — at least to a point.
“With an advanced world model, an AI could develop a personal understanding of whatever scenario it’s placed in,” he said, “and start to reason out possible solutions.”
(Excerpt) Read more at techcrunch.com ...
Very interesting article. Seriously doubt that our world can be simulated to the point where every action doesn’t turn into some kind of butterfly effect.
photo of Donald Trump shooting many drones down
https://x.com/i/grok
Our world can’t be simulated using the current storage and computing limitations. But we can restrict the variables to be able to predict 99.99% of the time.
What I mean is - predict where the basketball will bounce - you need force, direction, ground even-ness, wind, other vibrations, moisture etc. - not too many.
But predicting larger scale - nope, not right now
But if our world is a simulation, then there exists a computer capable of runnig our world’s simulation.
So all we need to do is figure out how to hack into THAT computer, and then we can create any kind of world we want.
Reality is that “all models are wrong, though some are useful” is a rule written by George Box in 1976...To repeat: ALL models are wrong, though some are useful. That should be the underlying rule of modelling, though it’s often ignored.
I’ve been writing and using models since my first on a vacuum tube analog computer, a Pace 231R, in 1967. I can prove that all been wrong from the data. I’ve also found that some had been useful. Unfortunately, many were mis-used, too. That certainly is the case with AI.
I look at it as trying to break the speed of light. Even at 99.99%, almost infinite energy and mass of whatever is trying to get to that speed.
for such a device, it would have to have the power of the sun to draw on and storage and process would require something as large to exist on as the world it was trying to sim....IMHO.
To take in and process every variable to me is impossible. Not to mention the amount of programmers required to make the damned thing work.
if our world is a simulation, then we are pieces of code in the simulation.
We can’t hack into ourselves.
No doubt about it. Once the butterfly effect is important no idea what will happen. Suppress the tea party and you get MAGA and then Dark MAGA.
So, Dark Maga is the same antipathy that killed the Tea Party?
When it becomes a part of the ecosystem it has to simulate itself simulating itself simulating itself simulating itself…
You cannot prove the outcome. This is Godels theorem in a nutshell.
Thanks. Any way to use it without having an X account?
No. It is the unpredictable result of suppressing the tea party. Dark maga is MAGA joined by Musk et al which put paid to WEF style globomoism.
That is what Schwab and Gates tried to do. It was lunacy starting from the solution to mankind’s problems are vaxes and cow fart mitigation.
almost Newtonian.
Are you sure. It really depends on whether we have access to the similation code and a way to modify it. Or at least influence it.
Imagine an AI. That AI has the ability to write programs, and CHATGPT and many other AI programs can indeed write programs. Now if they have access to a shell program, then they can run the programs they write. And if the AI has access to it's own code, then it can make modifications to its own code.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.