New Figure humanoid




Wow! Just “Wow”!!

Thanks for sharing; I had my doubts about how quickly things in this field would move but I may have to recalculate.



“Watch MSNBC and Fox for the next two hours and be ready to discuss what’s going on right now?”
“Now that you’ve watched, what do you think?”

d fb

I was suitably impressed, how they managed to marry speech and the more traditional ‘deep learning.’ I have a special interest in the technology based on my experience in programming. I have never worked on AI beyond tinkering with early ‘expert systems’ that proved to be duds to the point that AI was declared dead on arrival! My programming background is only in heuristics but some incidents shed light on how the brain works, at least on how it works from the perspective of a computer programmer. If not interested, skip to the next post.

First, there are two brain functions, the boolean reasoning brain that does math, lacking or underdeveloped in all but humans, and the subconscious or lizard brain, that runs most of life forms. From this programmer’s perspective, this subconscious stores vast amounts of data, all the experiences since birth, and can recall any bit of it by some mechanism of pattern matching. Until the invention and development of neural networks we had no means of replicating the pattern matching capacity of the lizard brain. The Turing / von Neumann computer, by contrast, is a boolean architecture, if this then that. One curious factoid, it’s all binary, yes/no, 1/0, hole/no hole. The difference is only in the complexity of the system, the bigger the system, the more complex the system, the more ‘intelligent’ it becomes (whatever intelligence means).

Second, by necessity the brain can only store symbolic data, you can’t have a stone pyramid in your brain, not even a grain of sand. Language is the ability to create transmissible symbols of both real and abstract things. Language vasty expands the things brains can store, the meme pyramid in lieu of the stone pyramid. How difficult is to describe a face or a wine? It’s much easier using a picture or a sip of the wine. To create humanlike AI we need both ‘deep learning’ and ‘large language models.’

Third, ‘deep learning’ and ‘large language models’ are actually the same thing, the only difference is the input into the neural networks. Deep learning sees the world through sensory devices, cameras (pixels) and microphones (frequencies, etc.) while large language models see the world through memes (words).

Fourth, the level of intelligence depends mostly on the capacity and speed of the underlying neural networks which, in turn, depend on the capacity and speed of the hardware they run on. Size matters! Need proof? How large and complex are genomes?

One copy of the human genome consists of approximately 3 billion base pairs of DNA, which are distributed across 23 chromosomes. (9 zeros)

The average adult male has around 36 trillion cells in their body, while average adult females have 28 trillion, researchers have found. (12 zeros)

Total, over 21 zeros

Size matters!

The Captain


I seriously have my doubts about ANY AI/computer generated voice system going “eh…” in the middle of a sentence. That makes me immediately suspicious of this demonstration.

(Right during the explanation of why robot handed human an apple.)

edit - ((Ha, they address it right off the bat, but I am still suspicious of the pattern of speech and the way the robot ‘drops’ the apple into the outheld hand. That is a level of predictive analysis and human muscle memory that is way difficult to emulate…))

1 Like

You don’t remember the demos a few years back of an AI voice calling to make dinner reservations? It was full of moments of speech like that because it learned to speak from actual humans.