About the title of this page: AI gurus were claiming super-human AI intelligence was just around the corner 70 years ago. 50 years ago, a lot of school kids were yelling “LOOK MOM!” when they got their first handheld calculator. The kids truly believed that computers would soon surpass humans in intelligence (those calculators already had super-human abilities for calculating numbers, so everyone knew it was just a (short) matter of time before computers had super-duper super-human intelligence (AGI)). The AI gurus were just pretending to believe.

This page is an overview of the ZiptieAI.com pages that show that the very idea of AI or AGI (super-duper AI) having one iota of intelligence is nonsense (AI/AGI are just digital (binary) computation). This is important so that you

Get an intrinisic feeling for what problems AI can solve (realistically, reliably).
Won’t fall for the rampant AI hype.

TOC

(1) 5 weeks in Wuxi: AI concepts from top to bottom. I spent most of that time rewriting some key ZiptieAI concepts pages. I did this without an internet connection (the result is a lot more of the old-style pre-AI original thinking (and some pretty rough diagrams done in MS Paint)).
(2) The AI reality (NO WIRES). The biggest lesson from those pages: AI “neural networks” (NN) have wires. Top to bottom. Real NNs are wired.
(3) The AI con game and the real deal. No wires means no intelligence. Claims of AI/AGI having human-like intelligence are a con game. But AI has revolutionary capabilities.
(4) In the future GI (genuine intelligence) will be wired (and wild beyond belief). The wiring in your brain will be the blueprint for GI. GI will be hosted on man-made substrates that replicate the architecture of core parts of the human brain.

(1) 5 weeks in Wuxi: AI concepts from top to bottom

I spent 5 weeks in Wuxi (China) in Sep/Oct 2026 on “vacation”. No AI, no search engines, just me and my PC. Just like back in the good ol’ days before the internet. I spent most of that time updating or adding the pages described below that describe AI

from top (the enterprise (3c))
to bottom (the simplest Tiny LLM demos (2b))

xxx

TOC

(1.1) 3c Enterprise (updated)
(1.2) Core Foundry Enterprise AI Concepts (rewritten)
(1.3) Core LLM Concepts (added)

(1.1) 3c Enterprise (updated)

The top-level webpage about all things Palantir (Foundry).

xxx

(1.2) Core Foundry Enterprise AI Concepts (rewritten)

See section 3 “LLMs = pattern recognition”.

xxx

(1.3) Core LLM Concepts (added)

xxx

(2) The AI reality (NO WIRES)

Section “(1.3) Core LLM Concepts” (above) described the ZAI page that shows how to run the TINY LLM demo on a GPU or on a CPU. That “puts the nail in the coffin” of the idea that AI has one iota of human intelligence. A CPU is not a NN (neural network) – there are no wires between nodes (this is also true for GPUs, but a CPU makes it more obvious).

(left) Matrix math and (right) a real (wired) NN.
xxx xxx

To understand the core of computational AI, you need to study the code for the simplest transformer (TF).

xxx

The following ZAI pages describe the D5 TINY TF demo (the bottom level AI demo).

(2.1) D5 Tiny TF
(2.2) D5 tiny TF algorithm details

(2.1) 2.3.6.1 D5 Tiny TF

The simplest possible hands on tiny TF demo. Anyone who wants to understand how AI really works should study and run this demo.

xxx

(2.2) 2.3.6.1b D5 tiny TF algorithm details

The Tiny TF demo code details explained (tech conversations with GPT).

xxx

(3) The AI con game and the real deal

(WIP)

TOC

3.1 The AI con game. Claims that AI has human-like intelligence are nonsense. AI consists of binary computational algorithms and the binary HW to run those algorithms. The human brain is vastly more than that.
3.2 The real value of AI. But non-intelligent AI SW/HW are revolutionary when used properly. AI is changing our world.

3.1 The AI con game

Jeff Dean and his team left Google recently ostensibly to create a new startup. I believe they were asked to leave. After a decade of investment, Waymo cars still can’t drive themselves. And they never will. Musk has yet to deliver on FSD. He’s now busy planning to colonize Mars (or the moon? which one is it this week?) with his “star” ships. Musk knows that what he is saying is nonsense, but its good for business (his business, not yours; billionaires are smart people).
An unsuccessful application of AI (you need real intelligence to drive a car).
xxx

The following diagram compares AI to brain functions. The diagram only confuses me.
Diagram from the blog “BestPartner”, one of my favorite blogs.
xxx

The “brain” in the above diagram was put there not to explain, but to mislead. Modern AI is no more intelligent than the AI from 70 years ago, but the con game is still the same. The following is from the ZAI webpage “Hack” section “5 TF GPU-based algorithms are 100% deterministic”.

The AI simulation of intelligence started to seem intelligent only recently after the computational hardware reached the performance level required to (1) generate responses at least as fast as a human could read them and (2) with a level of accuracy that (at least initially) convinced users that AI had some level of genuine intelligence. But AI has no intelligence. AI algorithms and the circuitry they run on have little similarity to human brain neurons. You could run the most sophisticated LLMs on GPUs based on electro-mechanical relays. It might require its own power plant and take years to answer a prompt, but it is theoretically possible.

Electro-mechanical relay
drones

You could probably also construct a purely mechanical (no electric whatsoever) AI device that ran on water power. It would accomplish the same thing as the EM relay (although requiring even more time, power and space). The point is, the device would be 100% mechanical, which is basically what a GPU is (in modern GPUs, electricity is only used as the medium/mechanism to perform the binary logic operations). Nothing more. Does anyone really believe that a water-powered mechanical AI device that produces the same results as GPU-based AI device would be intelligent?

3.2 The real value of AI

AI consists of the SW algorithms and the HW that can run AI algorithms at the speed required for human interaction (in the future the speed requirements will be increasingly determined not by human users, but by other AI devices). AI can not perform tasks that require intelligence, but it can perform many tasks that normally require a lot of human labor and intelligence (and do this much faster than humans). AI can also do many task that are impossible for humans. That is the real value of AI.

And, in case you did not notice, it is in essence not much different from the value we have been getting from electronic computers for past 80 years (AI simply raised the level of functionality and performance).

This is a ZAI version of the BestPartner diagram. The functionality shown below is not intelligent, but it is drastically changing our world (you can’t use it to safely drive cars or control robots in the home, but you can use it to give voice commands, ask for directions, help you (very slowly) park your car, etc).
xxx

(4) In the future GI (genuine intelligence) will be wired (and wild beyond belief)

(work in progress)

This is the “AI” topic that interests me most: Genuine intelligence (GI) hosted on man-made substrates that mimic the architecture of the human brain. GI is my own term; I have not read anything about this topic anywhere but on ZiptieAI.com (I wrote about this a year ago, but then removed the page). The ultimate moral dilemma and pandora’s box. It will come (and it will be the ultimate pandora’s box).

SUMMARY (the following is based on my own ideas (guesses, assumptions, etc); I hesitate to discuss such topics with token generators).

DNA directed the creation of your body.
- The driving force directly this construction was EM forces (on a molecular level).
Its not too far fetched to suggest that EM (with crosstalk) in your brain (caused by electro-chemical signals) is the source of consciousness and intelligence.
- For GI it should also be on the EM fields in the spaces between the neuron wiring.
Replicating the process of using DNA to create the brain is impossible for now.
- It would be better if it remained so.
- But it wont.
But replicating the EM in the brain will be possible much sooner.
- This will be the ultimate pandoras box (it will happen). Nuclear weapons and “gain-of-function” bioweapons will be child’s play by comparison.

WORKING NOTES

There are 4 fundamental forces in nature

Strong force (in nucleus)
Weak force (in nucleus)
Gravity
EM (electromagnetism)

Human intelligence and (most importantly) conciousness can only come from EM.

Otherwise: You have to add a 5th mysterious fundamental force.

DNA self replicates.

The most complex process in the universe.
That replication depends on EM interaction.

The source of consciousness, senses, and intelligence in the human brain

The neuron wiring.
The EM fields between wiring (and the crosstalk).
Crosstalk happens because
- The EM energy is not only in the wiring, but in the space around the wire.
- Space is not empty (even in a vacuum).

You can see perturbations in space (light beams) because

Detectors in your eyes only react to such EM perturbations.
intelligence (consciousness) = EM space perturbations in your brain.

Space is not empty (the vacuum is not a vacuum).

We cant sense static air pressure (10 tons / sq meter), but that does not mean its not there.
We simply are not designed to detect the massive foundation that supports our world.
Space is like an ocean
- We cant see the ocean (like we cant see air pressure). But the ocean is definitely there.
- Waves in that ocean are all we can detect.
- Those waves are also the basis of our consciousness and intelligence.
- Those waves are of very low power (compared to what the ocean can sustain).
  - Just like chemical reactions release an extrememly tiny fraction of the energy stored in the mass of the reactants.

GI will be much different than bio intelligence:

None of the bio restrictions in brain size and power levels
- You could power GI with nuclear power
  - (bio brains are limited to what can be extracted from food sources)
- GI brain signals will travel ~10,000,000 times faster (lightspeed = 300,000,000 m/sec compared to a bio signal’s ~30m/s).
  - Therefore 1 GI brain sec would = 100 human brain days (they would think vastly faster).
- Easily mass produced.
- GI brain can be designed to have any temperament, to be in constant pain, whatever.
- GI brain can have native interfaces for any types of sensors.

The ultimate pandora’s box.

26.1006 (v1 26.1001)