The essence of the expected breakthrough is the resolution of the Moravec paradox, an indestructible wall blocking the path to humanoid Strong-AI (AGI), and then to Super-AI.
The logic here is simple. To achieve the AGI level, an intelligent agent must have a body (it is the one that will “live”, adapting to the external environment and interacting with it and its own kind).
However, on the way to the corporealization of AI stands Moravec’s paradox, according to which. Controlling low-level sensorimotor operations (the body’s operations) requires enormous computational resources, even greater than the control of high-cognitive processes (the complex mental operations of the brain).
Those without a humanoid body (robot – android) there will be no AGI, and the android does not have enough computing resources to “live”.
What do they promise in March
Bernt Bornich , the CEO of 1X (which OpenAI is betting on in Android robotics), announced that in March it looks like it will be proven that the Moravec paradox limitation was false, and it arose only due to a banal lack of data.
We are most likely talking about the release of the android NEO – the “brother” of Eve (android EVE). NEO will be superior to its “sister” in many ways and, in addition, will move on legs rather than on wheels.
The January demonstration of Eve was so amazing that the editors of IEEE Spectrum even conducted their own investigation into whether people were hiding under the android suits. But everything turned out to be clean – you can’t dig into it.
And this is a detailed analysis of the demonstration:
An Android feature from 1X in an engine developed by OpenAI. This neuron, like the human mind, works by learning from data coming from vision and generating actions (control of movement, arms, grips, torso and head) with a frequency of 10 Hz.
The “base model” understands a wide range of physical behavior, from cleaning the house to picking up items in a warehouse to social interaction with humans and other robots. But the most important thing is that new skills appear in the android in just a few minutes of data collection and training – by observing how people do it.
For skeptics who doubt the impending refutation of Moravec’s paradox (even if OpenAI’s ears are visible behind it), I recommend the announcement made the day before yesterday by Ted Xiao (Senior Research Scientist at Google DeepMind Robotics).
Ted wrote this:
“There will be 3-4 massive news coming out In the next weeks that will rock the robotics and artificial intelligence space. Adjust your timelines, it will be a crazy 2024.”
James Darpinian writes about the same thing in his best review of the TOP 20 androids in the world today:
The Holy Grail—a walking, talking, helpful science fiction android—is suddenly within reach.
So, we are waiting for the robot revolution in March.
