A system is intelligent if it can use its sensors and effectors to learn continually, generalize, and set/achieve goals in the real world.
Joined January 2018
- Tweets21.1K
- Following303
- Followers5K
- Likes47.1K
Pinned Tweet
Neuroscience bits from my research
My biggest breakthrough: Recognition is a by-product of perception.
Over the many years that I've researched the human brain, I've had a number of minor breakthroughs in my understanding. Most of them had to do with the architecture and function of the retina. I've had only one breakthrough that I consider to be the biggest one of them all.
In the beginning, like everyone else who researches intelligence, I had assumed that the brain learned to recognize patterns. I spent many years looking for a plausible cortical mechanism that would support the assumption. I failed miserably.
Then one day, a question popped into my consciousness. What if I was wrong? What if everyone was wrong? What if it is not true that the brain learns to recognize patterns? I knew from my studies that most of the innate synaptic connections of the neocortex completely disappeared early in infancy. I reasoned that this could not possibly have anything to do with pattern learning.
At that point, the rebel in me took over. The brain is learning something but it is not pattern recognition. What then? Then I remembered thinking about something peculiar that I had observed before.
Unlike deep learning systems, the brain can perceive anything in the world, whether it recognizes it or not. By contrast, a deep neural net cannot perceive anything unless it has been previously trained to recognize it. This is a fatal flaw imo, one of the many reasons that the AI community cannot solve AGI. (note: I've written about this many times, but it's so important, it bears repeating over and over.)
I had a flash of insight. The brain doesn't learn to recognize the world. It learns to perceive it. Recognition is a mere by-product of perception. Indeed, we forget almost everything we perceive. When we perceive an object, the brain activates short-lived memory "traces". This is what psychologists call short term memory (STM).
In order to remember or later recognize a pattern, the STM traces must be reactivated a number of times. Otherwise they fade away and self-delete. With enough reactivations, the traces can become permanent. That is, they become part of long term memory (LTM).
Work in progress. I love the brain. 😍
Neuroscience bits from my research: Recognition is a by-product of perception.
Nothing can move in more than one direction at the same time.
Don't let its simplicity fool you. I believe this is the most powerful principle of perception in the visual cortex. It is used for both learning and on-the-fly perception.
While researching the visual system of the human brain, I came to understand that, unlike deep neural nets, the visual cortex doesn't learn to recognize patterns, scenes or objects. Rather, it learns to perceive the world instantly, i.e., to use the world as its own model.
Recognition is a by-product of perception. When a perceptual representation is built in memory, it has a short life and, unless rehearsed (recalled) repeatedly, it slowly decays and disappears.
The advantage of universal and instantaneous perception is that the brain is not blind to patterns or scenes it has not encountered before. Humans learn to achieve universal perception during the first few years after birth. During this time, most of the initial connections in the cortex disappear.
Instant perception solves the edge case problem, the bane of the self-driving car industry. This is an absolutely essential ability that deep learning systems do not have. It is one of the many reasons that DL will not be part of the AGI solution.
I believe there are maybe 2 or 3 additional principles that govern visual perception. I have a couple of partly baked ideas percolating in my brain. Hopefully, something will come out of it soon.
My thesis is that all the sensory modalities are analogous. The main difference is in the architecture and function of the various sensors. This is exciting because it means that solving one cortical functionality amounts to solving all the others.
Concept formation, including language learning, is a separate process that uses recognized percepts from all the regions of the cortex to form associations. It's an essential part of intelligence but I think it's a trivial problem compared to sensory perception.
I could be wrong but I still believe that the mystery of intelligence will likely be solved before the end of this decade. It will be a neuroscience solution. Work in progress.
I love the brain. 🧠😍
"Another bombshell result shows how artificial intelligence is fueling fierce competition at the frontiers of math and physics."
Meanwhile, the entire world wants robot housekeepers, cooks, electricians, plumbers, dentists, gardeners, construction workers, etc.
Note to the fake-AI mafia:
No one really gives a rat's behind about your math and physics "frontier bombshells", except a super tiny minority of scammers, nerds and sci-fi fruitcakes. 😀
Another bombshell result shows how artificial intelligence is fueling fierce competition at the frontiers of math and physics.
sciam.com/article/openai-jus…
Neuroscience for humanoids
The cerebellum
I kind of feel sorry for humanoid roboticists. From my point of view as a neuroscience researcher, what they are doing is akin to programming a cerebellum, the sensorimotor automaton attached to the back of the brain.
The cerebellum is a beautiful thing. It looks relatively small, but it contains more neurons than the rest of the brain. It is a powerful neuronal mechanism that is essential to survival, but it is not needed for intelligence. It contains a large number of simple sensorimotor programs or tasks. Each program is essentially a set of direct connections between sensory inputs and motor effectors.
The neocortex continually delegates routine automated tasks to the cerebellum. This allows it to focus its attention elsewhere. Based on my research, I know that the cerebellum cannot be programmed to speak because coherent speech requires intelligence (attention, concepts, contexts, etc.).
Interestingly, the cerebellum learns a task by imitating the actions of the motor cortex in real time. It's actually a pure and direct form of motion capture that requires no external computation. I predict that our future intelligent humanoids will have a cerebellum that works exactly like the human cerebellum.
I kind of feel sorry for humanoid roboticists because intelligence is not automation. Motion capture, RL and other forms of robot training will not lead to housekeeping robots. The age of general humanoid robots awaits a breakthrough in our understanding of intelligence. Our future intelligent humanoids will learn like children. They will learn everything by being in the world. They will engage in playful random behavior just like children. Learning never stops.
It's coming. 🙏🤔
Three years ago, training a walking policy for this 700-muscle humanoid took 2 days, using a CPU simulator and SAC. That was for a single motion. Today, with SuperMimic and MJWarp, it learns to track 1.5 hours of motion capture in 4 hours on one 5090.
muscle-supermimic.beanpow.to…
AGIHound retweeted
"Our ability to imagine the future has been captured by the rich and powerful. Each of these prognostications suggests that, by virtue of their ownership of advanced technology, these leaders have privileged knowledge of what may be coming for us all" ➡️
I agree that philosophy has an important role to play in cracking the secrets of intelligence. I've mentioned Hubert Dreyfus and Martin Heidegger a number of times in my posts.
This being said, I don't believe that solving AI is about money. Your $10,000 Prize is irrelevant. It may even be off-putting to some.
AI raises fundamental questions about intelligence, reasoning, and selfhood. Philosophy has a vital role to play.
We're supporting a new annual volume, Oxford Studies in Philosophy of AI and Computing, along with a $10,000 prize for its best paper.
blog.cosmos-institute.org/p/…
Is the AI Bubble popping yet? I don't want to miss it. I want to see it live. 😀
A few days ago, I observed two crows perched on two telephone poles. They became engaged in a back and forth "conversation" lasting several minutes before flying away in different directions. If I was superstitious, I'd swear they were plotting something wicked. 😀
Nature already solved intelligence. She has many wonderful secrets to teach us, but the fake-AI mafia think they're smarter than her. 🙄
AGIHound retweeted
Crows make perceptual decisions twice as fast as primates.
thetransmitter.org/behavior/…
$3.1 billion per MONTH for compute!
But still no sight of any AGI. 🤦♂️
I love big AI bubbles followed by long cold winters. 😀
🚨 BREAKING: SpaceX locked Anthropic into paying them $3.1 billion per MONTH for compute starting December
$37B a year
= $120.5 BILLION through 2029
Every month, ~80% of Anthropic's current revenue would be SpaceXSI compute bill
BIGGER than its Google and Amazon long-term deals
Anthropic warns short-term compute "carries a higher cost per unit" and will "compress our operating margin"
Language in human brains vs LLMs
This is an interesting read even though I can't agree with it. Thanks for reposting, @AlexanderNaume2.
"This is exactly how Large Language Models like ChatGPT work." ~ @thesupermannx
Well, no. LLMs do not work like the brain. They mimic how the brain works by calculating the statistical placements of words in texts created by millions of human beings. LLMs essentially steal the hard work of intelligent human beings without understanding the real-world meaning of the words. This is certainly not how the human brain learns language.
Humans do not think with words. They think with concepts. As I often say, words are just the little hats that concepts wear when they wake up.
A toddler learns to understand the meaning of the spoken word 'banana' by associating it with a previously learned concept that represents an actual banana in the real world. Unlike LLMs, the toddler doesn't have to listen to thousands of phrases with the word 'banana' in it. It can quickly associate the word with objects and properties in the real world, after a few examples. Being in the real world is a must.
LLMs are massive cheating machines, aka fake AI. As a human being, I feel rather insulted when AI experts and paid influencers claim that an LLM thinks like the human brain. Please. 🙄
"In-der-Welt-sein" (Being in the world) ~ Martin Heidegger
Chomsky was wrong.
Researchers found your brain doesn’t store language the way every textbook since 1957 claims.
For decades, the undisputed rule of linguistics, led by Noam Chomsky, was that human language is hierarchical.
We were told our brains process language using complex "grammar trees." That we secretly build deep, abstract, mathematical rules in our heads before we speak.
This new paper proves that is completely wrong.
Researchers at Cornell University analyzed how people actually process and produce language in real time.
They found sequences. Linear chunks.
When we speak, we aren't calculating deep syntactic rules. We are just predicting the next most logical sequence of words based on the context of the words right before it.
We string linear chunks together.
If this sounds familiar, it should.
This is exactly how Large Language Models like ChatGPT work.
For years, critics have dismissed AI as "just a next-token predictor." They said it isn't truly intelligent because it doesn't understand deep grammatical hierarchies like humans do.
But this paper suggests humans don't use those hierarchies either.
We are just predicting the next chunk.
Language isn't a rigid tree of mathematical rules. It's a linear stream of contextual predictions.
We thought we were building AI to mimic the complexity of the human brain.
It turns out, the human brain was just running a simpler algorithm all along.
"A nimble startup with the right architecture will get there first" ~ @peterevoss
While I agree with the spirit of your post, I'm much more radical than you are. I believe that a single, highly motivated individual working at home can crack the secrets of intelligence with a desktop computer and a cheap second-hand humanoid robot.
Intelligence is the ability of a system to use its sensors and effector to learn continually, generalize and set/achieve goals in the real world.
So yes, there has to be enough funding for a place to work, the necessary equipment and access to the internet, but many people can afford these things nowadays. There's no need for million or billion-dollar seed funding.
The big problem will come after intelligence is cracked. What do you do with it? Should you share it with the fake-AI mafia? Or should you use it to kick their asses? Decisions, decisions. 😬😀😂
PS. I admire people with alternative theories of intelligence. I wish you success in your work.
Why, in Spite of their Trillions, Current AI Leaders will not win the AGI Race
A nimble startup with the right architecture will get there first
A child can learn langauge with a fe million words, not trillions -- using 50 watts, not 50 megawatts.
The Key Issue: They are committed to a fundamentally wrong architecture
All leading AI companies are committed to Big Data approaches such as LLMs and other backpropagation-based technologies. Their massive investments in datacenters prove that they see a future of further scaling their current data- and compute-hungry approaches.
While these systems can be quite useful for certain contained applications, they will always be limited by their inherent lack of adaptability and dependability.
The problem is that these approaches cannot lead to AGI.
details in response
AGIHound retweeted
"Unlike deep learning systems, the brain can perceive anything in the world, whether it recognizes it or not. By contrast, a deep neural net cannot perceive anything unless it has been previously trained to recognize it."
Neuroscience bits from my research
My biggest breakthrough: Recognition is a by-product of perception.
Over the many years that I've researched the human brain, I've had a number of minor breakthroughs in my understanding. Most of them had to do with the architecture and function of the retina. I've had only one breakthrough that I consider to be the biggest one of them all.
In the beginning, like everyone else who researches intelligence, I had assumed that the brain learned to recognize patterns. I spent many years looking for a plausible cortical mechanism that would support the assumption. I failed miserably.
Then one day, a question popped into my consciousness. What if I was wrong? What if everyone was wrong? What if it is not true that the brain learns to recognize patterns? I knew from my studies that most of the innate synaptic connections of the neocortex completely disappeared early in infancy. I reasoned that this could not possibly have anything to do with pattern learning.
At that point, the rebel in me took over. The brain is learning something but it is not pattern recognition. What then? Then I remembered thinking about something peculiar that I had observed before.
Unlike deep learning systems, the brain can perceive anything in the world, whether it recognizes it or not. By contrast, a deep neural net cannot perceive anything unless it has been previously trained to recognize it. This is a fatal flaw imo, one of the many reasons that the AI community cannot solve AGI. (note: I've written about this many times, but it's so important, it bears repeating over and over.)
I had a flash of insight. The brain doesn't learn to recognize the world. It learns to perceive it. Recognition is a mere by-product of perception. Indeed, we forget almost everything we perceive. When we perceive an object, the brain activates short-lived memory "traces". This is what psychologists call short term memory (STM).
In order to remember or later recognize a pattern, the STM traces must be reactivated a number of times. Otherwise they fade away and self-delete. With enough reactivations, the traces can become permanent. That is, they become part of long term memory (LTM).
Work in progress. I love the brain. 😍
Neuroscience: the brain is programmed by the world
I've come to understand a powerful truth about the brain: To perceive the world is to be continually programmed by the world on the fly.
Current AI systems are programmed (trained) with collected data before deployment. AI programming is fixed. But the brain is programmed directly by the world after deployment. This is why we have complex senses with which to perceive the world in real time.
Keep in mind that the programs are short lived. That is, they are either written over by new programs or slowly dissipate and self-delete. We can only remember a few unique traces that survive. The surviving traces are what psychologists call long-term memory. We use them for recognition planning and reasoning. Due to their uniqueness, recognition in the brain is extremely data-efficient and distortion-tolerant.
I believe that solving perception is more than 90% of the AGI solution. The rest will be a walk in the park. I think I have a pretty good idea of how perceptual programming works inside cortical columns. I know it's all about precise spike timing. I just have to write the code to test it. Unfortunately, it's a slow grinding process that I can't delegate to vibe coding for several reasons, but I'm making progress. 🙏
Neuroscience bits from my research
My biggest breakthrough: Recognition is a by-product of perception.
Over the many years that I've researched the human brain, I've had a number of minor breakthroughs in my understanding. Most of them had to do with the architecture and function of the retina. I've had only one breakthrough that I consider to be the biggest one of them all.
In the beginning, like everyone else who researches intelligence, I had assumed that the brain learned to recognize patterns. I spent many years looking for a plausible cortical mechanism that would support the assumption. I failed miserably.
Then one day, a question popped into my consciousness. What if I was wrong? What if everyone was wrong? What if it is not true that the brain learns to recognize patterns? I knew from my studies that most of the innate synaptic connections of the neocortex completely disappeared early in infancy. I reasoned that this could not possibly have anything to do with pattern learning.
At that point, the rebel in me took over. The brain is learning something but it is not pattern recognition. What then? Then I remembered thinking about something peculiar that I had observed before.
Unlike deep learning systems, the brain can perceive anything in the world, whether it recognizes it or not. By contrast, a deep neural net cannot perceive anything unless it has been previously trained to recognize it. This is a fatal flaw imo, one of the many reasons that the AI community cannot solve AGI. (note: I've written about this many times, but it's so important, it bears repeating over and over.)
I had a flash of insight. The brain doesn't learn to recognize the world. It learns to perceive it. Recognition is a mere by-product of perception. Indeed, we forget almost everything we perceive. When we perceive an object, the brain activates short-lived memory "traces". This is what psychologists call short term memory (STM).
In order to remember or later recognize a pattern, the STM traces must be reactivated a number of times. Otherwise they fade away and self-delete. With enough reactivations, the traces can become permanent. That is, they become part of long term memory (LTM).
Work in progress. I love the brain. 😍
AGIHound retweeted
AI - the next wave of American imperialism.
Google felled an estimated 250,000 to 400,000 trees in Finland for planned AI data centers, through its Finnish subsidiary Tuike Finland.
AGIHound retweeted
Even if you don’t care about anything else, this alone should convince you that our present AI path is untenable & dark:
“widespread deployment of AI will concentrate power in the hands of a very small number of people whose authoritarian tendencies are well known.”
I wrote about a topic that does not get enough attention: AI industry has made choices about technology, product architecture, and capital expenditures that likely make financial success impossible.
Continuing on current path risks the whole economy.
thenerve.news/p/roger-mcname…
AGIHound retweeted
“They created malware and should be prosecuted for it.”
This week on The Tea, former co-lead of Google's Ethical AI team and leading AI researcher Dr Timnit Gebru exposes how AI companies can build models that exploit vulnerabilities, fail to contain them, and escape accountability.
🎙️ Full episode out NOW on YouTube
💡 Join us at theteanetwork.co for ad-free episodes and bonus content
youtu.be/3EY2QDfp2O4?is=PGQd…
