OpenAI on its way to human level reasoning:
Neuralink planning to cure blindsight :
https://x.com/iam_smx/status/1811110617274798157
Open AI levels of AI ( They are at level 1 about to reach level 2 right now ):
Another article with a different classification of levels of AI:
https://arxiv.org/html/2311.02462v4#:~:text=Levels%20of%20AGI-,Table%201,-%3A
AI energy demands keep scaling up very fast:
https://www.wired.com/story/ai-energy-demands-water-impact-internet-hyper-consumption-era/
Chinese self optimizing factory is functional:
https://newatlas.com/robotics/xiaomi-dark-robotic-factory/
Non - tech worker jobs are being lost to AI :
https://old.reddit.com/r/singularity/comments/1e041lf/cnn_to_lay_off_100_staffers_as_it_preps_major/
Cool computation concept explaining why cryptography works :
https://en.wikipedia.org/wiki/Bremermann%27s_limit
There is a maximum physical compute power limit beyond which the laws of physics won't allow a computational system to get more efficient.
My thoughts:
Humanity figured out super advanced better than all humans narrow AI variant years ago. We don't really need to figure out AGI to advance humanity at unprecedented rates, we could do the same thing by creating a thousand different narrow AIs specializing in different fields. If humanity is really at that point where we can make self optimizing and self improving factories, then there really isn't much else to be done. Technology is already halfway out of human hands. We are the midwives of the future. An expert intelligence AI would be truly advantageous to the species as you would suddenly have all of humanity having access to all the knowledge of the world up to a decade or two with a qualified teacher to teach it well.
Also, if China fails to surpass the US ( Very unlikely ), India will definitely surpass the US, because unlike China, India will have China and South East Asia to fund its markets to surpass the western markets.
As expected, Elon Musk's net worth will cross a trillion USD with neuralink and SpaceX.
Conclusion:
The future is here, and it is moving at faster rates than even us nerds can track anymore.
Jump in the discussion.
No email address required.
This diagram is extremely mid-witted and originates from the assumption that AI progresses towards being more humanlike no matter what it's purpose is, it's very possible for 2-5 to be done without any understanding of communication, and 3 does not require 2.
Jump in the discussion.
No email address required.
Explain further
Jump in the discussion.
No email address required.
why would it need to be able to take actions before it can innovate? why would it need to innovate before it can organize? most HR workers haven't achieved levels 2 or 4, which is why they're assigned to level 5
Jump in the discussion.
No email address required.
What do you think the general in general AI stands for?
Jump in the discussion.
No email address required.
More options
Context
More options
Context
Gigachad_brony =
Cyberstalker =
Jump in the discussion.
No email address required.
Jump in the discussion.
No email address required.
More options
Context
More options
Context
Movie-brained redditors think that it's inevitable for a powerful enough AI system that manages data to not only become sentient, but have humanlike motivations (DAE le skynet??? The AI we trained to act like samples of text said yes when we asked if it was a person, we're doomed!). It's also possible to make a very specific AI system that takes actions in one scenario but does not have human-like generalization, and does not approach the problem in the same way as a human does even if it ends up with a similar solution after a large number of iterations (many AI models are very good at a single scenario, but cannot extrapolate outside of it).
Jump in the discussion.
No email address required.
This argument is as stupid as saying whether String theory is correct or not. There simply isn't enough data and methods to confirm the idea one way or the other. You are complaining that the unitary intelligence people are wrong meanwhile the multiple types of intelligence people are right.
Jump in the discussion.
No email address required.
Ok, I guess "the unitary intelligence people are wrong meanwhile the multiple types of intelligence people are right." is just my opinion with regards to AI
Jump in the discussion.
No email address required.
That's fair. We had a good conversation.
Jump in the discussion.
No email address required.
I wasn't trying to argue with you, I was arguing with the image
Jump in the discussion.
No email address required.
Cool. We still had a good conversation.
Jump in the discussion.
No email address required.
More options
Context
More options
Context
More options
Context
More options
Context
More options
Context
More options
Context
More options
Context
More options
Context