A system is intelligent if it can use its sensors and effectors to learn continually, generalize, and set/achieve goals in the real world.
- Tweets 21,226
- Following 300
- Followers 4,979
- Likes 47,708
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. 🧠😍
9
4
44
8,104
Physics for rebels
"When Mercury's orbit didn't match the models of Newtonian physics, astronomers assumed there must be another planet nearby throwing off the orbit.
They were wrong. They needed Einstein’s theory of relativity to rewrite their basic physics assumptions."
This is the propaganda but it is false. It is known that they are grave errors in Einstein's calculations for the precession of the perihelion of Mercury's orbit.
It will be found that the tiny anomaly in Mercury's perihelion precession is due to the dark matter that surrounds the Sun. As several researchers have shown, Mercury's orbit anomaly disappears completely by using a slightly smaller value for G when calculating Neptune's orbit. This is to be expected with dark matter.
The small but powerful political mafia that dominates the sciences will not allow Einstein to be wrong in the public's eye, but even Einstein knew by the end of his life that his theories were wrong because they were based on continuity, a bogus concept that requires infinity.
One year before he died, Einstein wrote to his friend Besso: "In that case, nothing remains of my entire castle in the air, gravitation theory included, [and of] the rest of modern physics." ~ from 'Subtle is the Lord' by Abraham Pais
We live in the age of bullshit, but not for too much longer. 😮
When Uranus's orbit didn't match the models of Newtonian physics, astronomers assumed there must be another planet nearby throwing off the orbit.
They soon discovered Neptune.
When Mercury's orbit didn't match the models of Newtonian physics, astronomers assumed there must be another planet nearby throwing off the orbit.
They were wrong. They needed Einstein’s theory of relativity to rewrite their basic physics assumptions.
@michael_nielsen explains why scientists often can’t tell whether an anomaly requires a minor correction or a scientific revolution.
115
AGIHound retweeted
AI that is "alive" will not be an answer machine.
It will arrive as a question machine.
@TrueAIHound
1
2
5
181
Neuroscience bits from my research
"I determined years ago that the human retina can detect edge movements in 10 orientations and 20 directions (2 per orientation)." ~ from the quoted post
Prediction
I know things about visual perception and learning, powerful things that I can't reveal at this time. For the record, I will predict that the following things will one day be understood as the foundational principles of the human visual system:
1. Precise timing of discrete sensory events (spikes).
2. Edge movements: 10 orientations and 20 directions.
3. Elimination of cortical connections that cause timing contradiction.
I believe that all the sensory modalities are analogous. That is to say, if you solve one modality, the others are also solved. The only difference will be in the design of the sensors. This understanding is much more powerful than it may seem.
I still stand by my estimate of when the secrets of intelligence will be cracked: no later than 2029. I could be wrong of course. It's just an estimate. 🙏
Work in progress. I love the brain. 😍
Neuroscience bits from my research: Perceptual learning in cortical columns
Every column has a boss
I've been making rapid progress in my understanding of the sensory cortex lately. I've come to understand that every cortical column is like a small team of players charged with a task, and that the team has a leader or boss. What this means is that the activity of the players must obey the boss's lead. (Note: I know how this works but I'm sorry that can't expand on it at this time.)
Edge movement orientations and directions
In the visual cortex, the task of a cortical column is tightly related to the timing of the orientation and direction signals arriving from the column's receptive field in the retina. I determined years ago that the human retina can detect edge movements in 10 orientations and 20 directions (2 per orientation).
Timing contradiction
My thesis is that perceptual learning in the visual cortex is strictly based on spike timing and the elimination of connections that cause timing contradictions. For example, an edge cannot move in more than one orientation or direction at the same time. I hypothesize that bad connections are flagged during wakeful hours and are fully disconnected during REM sleep.
Brain versus deep neural nets
Error back-propagation, gradients, connection weights and function optimization are not used by the brain. This means that the brain doesn't suffer from the problems that plague deep learning, such as overfitting, local optima or catastrophic forgetting.
Unlike deep neural nets, the brain learns continually in the real world. The world is the teacher. Also, the brain doesn't learn to recognize patterns. It took me many years to realize that I and everyone else were wrong about pattern recognition. The brain learns to perceive the world instantly. To perceive the world is to understand the temporal logic of the world in real time. This way, the brain can perceive any object, scene or pattern whether or not it recognizes it.
Short and long-term memory
During perception, cortical columns use inhibitory synapses to quickly gate arriving signals and form coherent percepts. The percepts are forgotten shortly after use (short-term memory) but, if rehearsed multiple times, surviving traces can become permanent and reused later for recognition. The surviving traces are called long-term memory in psychology.
Work in progress. I love the brain. 😍
ALT Source: https://www.allaboutvision.com/eye-care/eye-anatomy/eye-structure/retina/
2
1
10
460
I never did like that ugly thing. I saw it as an insult to human civilization and cultures. Bye.
2
11
354
Ok. Watching the AI bubble popping won't be enough. I want to see it followed by a long and cold AI winter.
I'm a traditional man. 😀
Anthropic's finances are deteriorating ahead of its IPO:
Anthropic reported +$16.96 billion in Q3 2026 revenue, up +47% QoQ.
However, quarterly revenue growth slowed sharply from +144% in Q2, as price reductions and contractual discounts offset revenue by ~$600 million.
Meanwhile, Anthropic's operating profitability lasted just ONE quarter as the company swung back to a GAAP operating loss of -$567 million in Q3, following operating income of +$974 million in Q2.
This comes as compute and infrastructure expenses surged +77% QoQ to $9.9 billion, accounting for 58% of revenue, up from 48% in Q2.
In effect, every $1 spent on computing infrastructure generated just $1.72 in revenue, down from $2.08 in Q2, marking the first quarterly deterioration in this key efficiency metric.
Overall, Anthropic's net loss reached -$23.3 billion in Q3, bringing its cumulative net loss for the first 9 months of 2026 to a staggering -$104 billion.
While most of the quarterly net loss came from non-cash valuation adjustments, the underlying operating business also slipped back into the red.
Excluding the $21.9 billion in non-cash valuation adjustments, Anthropic's Q3 net loss was -$1.4 billion.
Anthropic's trillion-dollar AI ambitions are increasingly colliding with the financial reality of an infrastructure race where computing costs are rising faster than revenue.
1
8
353
This is amazing. The hostility toward fake-AI scammer, Sam Altman, has reached the stars. Just take a look at the quotes to this post. It's brutal. 😬😮
My faith in humanity has been restored. 🙏
OpenAI CEO Sam Altman believes people owning RAM is wasteful, and that it belongs in data centers instead.
"From a global allocation perspective, actually, it’s quite terrible to have everyone have a bunch of the world’s limited supply of DRAM sitting in their own computer."
1
1
16
596
Dwarkesh Patel got mercilessly chastised for pushing his machine consciousness bullshit. Now he has a new hobby: let's bash the Roman empire. 🙄😀
The average ancient Roman was small and dumb by modern standards.
Even Caesar August was only about 5’6” tall.
@Oklahomaharper explains how Romans resembled the populations of the poorest countries in the world today due to disease exposure and bad nutrition.
3
2
16
964
My take is that AI is a BIBO technology: bullshit in, bullshit out. Before it can have any real impact on physics, it must have the ability to identify the massive amount of bullshit that has infested modern physics. 🤦♂️
That would be true intelligence, the kind that the fake-AI mafia has no clue how to make. Their fake AI just spreads the bullshit everywhere.😀
AI is progressing quickly and we can now begin to see its impact on physics
piped.video/watch?v=PUa5i_U8…
4
1
17
752
AGIHound retweeted
"Spending hundreds of billions of the people's money on mass surveillance is not a sign of confidence"
I have a theory about this strange new development.
The mass surveillance, fake-AI mafia want all your private data in their datacenters. Why? Is it because they need it to crack AGI? Of course not. My thesis is that they're afraid of AGI and are doing everything in their power to prevent it from being solved.
Spending hundreds of billions of the people's money on mass surveillance is not a sign of confidence. It's a sign that our wicked overlords are deathly afraid of something, but they have no idea where the threat will emerge. 😬😮
It's not that I'm complaining, mind you. They deserve to live in constant fear. 😀
1
4
188
I have a theory about this strange new development.
The mass surveillance, fake-AI mafia want all your private data in their datacenters. Why? Is it because they need it to crack AGI? Of course not. My thesis is that they're afraid of AGI and are doing everything in their power to prevent it from being solved.
Spending hundreds of billions of the people's money on mass surveillance is not a sign of confidence. It's a sign that our wicked overlords are deathly afraid of something, but they have no idea where the threat will emerge. 😬😮
It's not that I'm complaining, mind you. They deserve to live in constant fear. 😀
OpenAI CEO Sam Altman believes people owning RAM is wasteful, and that it belongs in data centers instead.
"From a global allocation perspective, actually, it’s quite terrible to have everyone have a bunch of the world’s limited supply of DRAM sitting in their own computer."
11
7
141
34,261
AGIHound retweeted
This is the flawed technology that we're destroying our natural environment for by racing to build so many data centres.
Watch these errors - errors like this are introduced into every system that AI is put into - healthcare, banking, education, military.
27
236
921
22,718
China is following the same AI path as the US. Unreliable agents are not the future. When the AI bubble pops, DeepSeek will crash hard just like its cousins in the US. Boom! 🍿🍺
1
1
8
344
AGIHound retweeted
Don't worry, there is no chance an LLM, based on silicon, will ever be conscious.
42
8
169
3,653
The end is near? Is the AI bubble about to burst?
I want to see Jensen Huang, Dario Amodei and Sam Altman crying like toddlers. I want to hear them quoting the Apocalypse: "Babylon the great has fallen!" 😀 🍿🍺
The World Will End If Nvidia Misses Earnings zerohedge.com/markets/world-…
4
25
863
