Agent Revolution

99 posts

Agent Revolution

Agent Revolution

@AgentRev

Engineering physicist, embedded dev, hardware exorcist, amateur diesel mechanic, AGI optimist

Canada Katılım Temmuz 2011
41 Takip Edilen40 Takipçiler
sbeve
sbeve@48psx·
@TrueAIHound I'm curious to hear your thoughts on how the brain solves the credit assignment problem.
English
2
0
0
3.1K
AGIHound
AGIHound@TrueAIHound·
Neuroscience I love this illusion because it's proof that human visual perception is based on edge movement detection by retinal ganglion cells (RGCs) in the retina. Deep learning experts have no clue what true vision is. It will revolutionize computer vision. I estimate that the data efficiency of pattern recognition in the brain is at least 2 orders of magnitude higher than DL. This alone will change everything. Work in progress.
Curiosity@CuriosityonX

These boxes are static. I repeat, these boxes are not moving. x.com/i/status/14760…

English
42
341
5.2K
240.9K
Agent Revolution
Agent Revolution@AgentRev·
@LuminousMonkey @TrueAIHound @LuminousMonkey The TBP team prefers sustainable growth rather than a boom and bust. They operate in a very unique and complex conceptual space; too many cooks would spoil the broth. Their diligence will bear fruit in due time.
English
0
0
1
29
LuminousMonkey
LuminousMonkey@LuminousMonkey·
@AgentRev @TrueAIHound It’s not my wheel house, but it feels like this is the correct way to go. Rather than throwing a huge amount of data into models. Wish you guys had that money that’s getting thrown into LLMs.
English
1
0
0
25
AGIHound
AGIHound@TrueAIHound·
New Year's resolution I research the brain because I want to understand intelligence. I don't get paid to do it and I don't ask for money. It's a labor of love and commitment. I've been working on it, on and off, for more than two decades, and I've made steady progress over the years. If, by some happenstance, the Gods smiled upon me and gave me the wisdom I need to crack the mystery of intelligence, I wouldn't sell it to anyone or any group for all the money in the world. Why? Because intelligence has existed for eons. It doesn't belong to anyone. I would only want to be compensated for my years of work. Every worker is worthy of his wages. Happy new year! May the new year bring pleasant surprises. Cheers! 🥂
AGIHound@TrueAIHound

Listening to AI experts talking about the brain is painful. As a neuroscience researcher, I was excited when I first saw that this interview was about the brain. But I stopped watching the video after 3 minutes. Marblestone admits that he's clueless about the brain but almost immediately starts talking about what types of loss and reward functions are used in the brain, while making comparisons with deep neural nets. Dear Lord. 🤦‍♂️ Please. The most important learning principle in the brain is the precise timing of discrete sensory events (spikes). Everything that is needed for perceptual learning is contained in the sensory spikes themselves and their relative timing. Visual learning relies on movements in the visual field. Essentially, the visual cortex learns by eliminating timing contradictions in the input spikes. Any allusion or comparison to loss or reward signals is a category error. Error feedback signals in the brain are only used for goal-directed motor learning. Deep neural nets (and function optimizers in general) are not even in the correct galaxy, let alone the correct ballpark of the AGI solution. AI experts fool themselves and others to believe that they're researching intelligence. They are not. They are computer automation experts. They should refrain from talking about neuroscience and stick to their own field. It's unnerving. 😬

English
5
0
27
1.4K
Agent Revolution
Agent Revolution@AgentRev·
@TrueAIHound @LuminousMonkey 8/8) The concept of reference frames can be ported to other sensory modalities, e.g. a specific guitar solo being part of a longer song. We'll have to wait and see where he's going with this, I wouldn't dismiss it just yet...
English
0
0
1
22
Agent Revolution
Agent Revolution@AgentRev·
@TrueAIHound @LuminousMonkey 7/8) but they started with 3D as a proof of concept. I agree with you about reference frames, they're more within the "Theory of Mind" realm than pure neuro. Although, I presume it's probably just semantics for associative memories applied to 3D data.
English
1
0
1
22
Agent Revolution
Agent Revolution@AgentRev·
@TrueAIHound @LuminousMonkey 4/5) Virtualizing neural spikes on silicon demands tremendous compute. The team's trying to figure out the abstract "machine code" that governs those spikes, to try and run it natively on transistors instead of simulated neurons.
English
2
0
1
38
AGIHound
AGIHound@TrueAIHound·
@LuminousMonkey I admire Hawkins and his team. Unlike AI experts, they are looking in the right place. Unfortunately, they are operating under certain assumptions imposed on them by their secular education. I wish them the best.
English
1
0
0
75
Agent Revolution
Agent Revolution@AgentRev·
@TrueAIHound The Euclidean 3D paradigm used by the Thousand Brains Project is just a placeholder to get things moving along while trying to figure out how the brain actually handles sparse codes of physical entities. Hawkins explains his reasoning at this timestamp: youtube.com/watch?v=lewMhw…
YouTube video
YouTube
English
0
0
1
36
AGIHound
AGIHound@TrueAIHound·
Neuroscience bits from my research. Why don't I think about 3D space in my research of the visual system of the human brain? It's because I believe it is abstract, a creation of the soul. I wrote earlier: "The beautiful and colorful 3D vista you think you perceive in the real world is mainly a creation of your soul." Some philosophers don't believe in the existence of space. Neither do I. The great German philosopher, Immanuel Kant, denied the existence of space. He once asked, "If space exists, where is it?" I agree with Kant. I also ask, if space exists, what is it made of? Kant was not alone in this view. Gottfried Leibniz, another great German philosopher, also denied the existence of space, absolute or otherwise. He wrote to his friend, Samuel Clarke, that "space is nothing other than an order of the existence of things [...]. And therefore, the fiction of a material finite universe, moving forward in an infinite empty space, cannot be admitted." The latest trends in the AI community are spatial intelligence and 3D world models. Strangely enough, while the current focus of my research is the visual system of the human brain, I never think of the 3D geometry of space. I use 2D geometry but only for the design of my simulated retina. In my view, the visual cortex doesn't care that the input signals (spikes) come from the retina or that they represent a 3D world. It cares only about their precise timing, luminance, colors and what I refer to as modes. Although the modes (20 in total) correspond to the orientations and directions of edge movement detectors in the retina, the visual cortex doesn't care about their provenance. So, when I see neuroscientists like Jeff Hawkins and his research team (Thousand Brains Project) or AI experts like Fei-Fei Li agonize over 3D geometry or reference frames, I feel somewhat sorry for them. Yes, I love the soul because it simplifies my work.
AGIHound tweet mediaAGIHound tweet media
AGIHound@TrueAIHound

Neuroscience prediction My position on consciousness is clear. It requires a knower and a known, a soul and a compatible brain. I'm so confident of this that I will predict that AGI cannot be solved by a soul denier. Only a soul believer can do it. Why do I say this? It's because a soul denier is forced to believe that the visual cortex perceives a 3D vista. It doesn't. The visual cortex is a discrete signal (spike) cataloguer. It has the same architecture as and works exactly like the auditory cortex or the somatosensory cortex. It relies on the precise timing of sensory spikes to make sense of the world. The only difference between vision and hearing is the sensors. The beautiful and colorful 3D vista you think you perceive in the real world is mainly a creation of your soul. Yes, your soul is a very powerful thing. It's what makes humans exceptional. I love the soul. 😍

English
4
0
22
1.1K
Parzival - ∞/89
Parzival - ∞/89@whyarethis·
@JacksonKernion Next step isn’t even RL. It is resonance computing with models that are continuously learn, self optimize, and grow without back propagation.
English
1
0
7
1.1K
Jackson Kernion
Jackson Kernion@JacksonKernion·
I'm trying to figure out what to care about next. I joined Anthropic 4+ years ago, motivated by the dream of building AGI. I was convinced from studying philosophy of mind that we're approaching sufficient scale and that anything that can be learned can be learned in an RL env.
English
152
45
1.9K
473.3K
JOverrated
JOverrated@JOverrated·
at this rate… this will be The Final Shape after Bungie fires all the devs:
English
54
332
2.7K
155.8K
SimplyChris クリス
SimplyChris クリス@simplychristv·
@BungieHelp I have 3 name for this emblem Aggressive Frame Aggressive Crafting Aggressive Weapons But i think you guys have a better name for it
English
12
0
8
5.2K
Bungie Server Status
Bungie Server Status@BNGServerStatus·
UPCOMING DESTINY 2 SERVER MAINTENANCE TIMELINE ❖ September 19th ❖ Start: 6 AM PDT (-7 UTC) ❖ Player Removal: 7:15 AM ❖ Expected log in: 10 AM ❖ Expected end: 12 PM More info: help.bungie.net/hc/articles/36…
English
225
112
1.3K
479.6K