Deepfakes will eventually be so real, it will take "Fake News" to new level.
They have advanced quickly already and a recent advance allows you to take a single picture and animate it into a living portrait.
'Deepfake' AI can turn the Mona Lisa into a convincing real person
A.I. can now turn the Mona Lisa into a real life person with just one picture.
A paper published by a Samsung artificial intelligence lab in Russia shows the ability to convincingly turn artworks and celebrity photographs into
moving images.
The technology raises the prospect of convincing "deep fakes", videos showing politicians and even people who have died saying things they never said,
created just with still images of their faces.
The paintings were animated using the facial expressions of a real person, while celebrity photographs of figures such as David Beckham were turned
into video using footage of them speaking on a different occasion.
The researchers, based at the Samsung AI Center in Moscow, also performed the effect on images of Marilyn Monroe and Albert Einstein, using the moving
facial features of other people which was then "mapped" onto their faces, a technique known as "puppeteering".
Other models have required a large set of images in order to create a fake video, but Samsung's method can do it with just one, though the effect is
more convincing if more images are used.
It's easy to imagine how advanced this will be 5 to 10 years from now.
Of course this poses a problem . We could get overrun with fake news and it will be hard to separate when the President says something or it's a fake.
You can get frame people and set them up. Here's the video, Mona Lisa starts at the 5:00 mark:
Also, this is just more evidence that the universe is being simulated and is a vast Quantum Computer as M.I.T. Professor Seth Lloyd says.
Scientist think that spacetime is a quantum error correcting code.
How Space and Time Could Be a Quantum Error-Correcting Code
In short, quantum error correction in a quantum computer is needed for the same reason error correction is needed for classical noisy channels. With
QC this deals with entanglement. So you need a lot of entangled spacetime to protect the information encoded on logical qubits from errors as its
being simulated. This could explain why space is so vast.
Scientist have also talked about quantum gates forming around black holes.
Spacetime Geometry near Rotating Black Holes Acts Like Quantum Computer, Physicist Says
According to a theoretical paper published in the Annals of Physics, by Dr. Ovidiu Racorean from the General Direction of Information Technology
in Bucharest, Romania, the geometry of spacetime around a rapidly spinning black hole (Kerr black hole) behaves like a quantum computer, and it can
encode photons with quantum messages.
It will be easy for AI to simulate different histories of the universe just based on information it processes online. You can back load the simulation
with our history and the simulated avatars would think the universe is billions of years old and there was a war for Independence. They can simulate a
history where America lost the war.
I think black holes are like computer servers that preserve information. This information can be rebooted and simulated over and over again.
Deepfakes will make simulations all too real. It will eventually be indistinguishable from "reality" if what we call "reality" isn't just a
simulation, within a simulation, within a simulation.......
edit on 23-5-2019 by neoholographic because: (no reason given)
But that doesn't look like Mona Lisa
Although yea, it does look like a regular real life person
Then again Mona Lisa isn't looking like herself either.
originally posted by: letni
But that doesn't look like Mona Lisa
Although yea, it does look like a regular real life person
Then again Mona Lisa isn't looking like herself either.
That's the point!
It's not supposed to look like Mona Lisa the painting. It's supposed to look like a regular, real life person and it does. Especially the middle one.
I don't think there was ever any question that this kind of thing could be done. The problem is that they're basically telling the public that in the
near future, we won't be able to believe anything we see or hear. That's kind of not good just from the standpoint of trust.
Knowing human nature, why would anyone trust anyone in a world where this is not only possible, but commonplace?
Yes, in the not so distant future this will challenge what is 'real'.
The number of applications this can and likely will be applied is only limited by one's imagination, for better or worse. Based on our current society
it will first and foremost be weaponized. I would say most definitely simulations as a tool of deception against enemies and by the MSM to further
dumb down and mass control the sheeple.
Fake news will be more like the 'Matrix' where most of the population won't be able to distinguish fake from real and/or will become part of the fake
world by choice which they won't even realize their choice is against their will.
On one hand this will create, form, and justify new age forensic labs to verify video before admissible in any court.
Some creative entrepreneur along with some talented techies will make a fortune and probably be fronted by the likes of an In-Q-Tel venture capitalist
along with their own certificate of Authenticity. Think profit & control foremost and maybe to better mankind a very distant third.
As some of us enter this new digital world in the near future will give a whole new meaning to the phrase 'Only the strong survive'.
When I read Nick Bostrom's "Simulation hypothesis" paper back in '03 and my software/networking background in future technology I knew one day this
would gain more traction and open science avenues to explore and experiment.
On the lighter side will certainly make for some awesome movies and video games seem and feel 'real'.
We are already at the point where you cannot trust anything. Here's one about fake
models. Here's one about making a movie from a still picture. Here's
one about faking voices. There is a youtube channel that covers a whole lot of
current AI research that's worth looking at. 2 minute papers