ChatGPT

[charliecoutas]

ChatGPT has helped me in several areas: epoxy resin problems, PIC programming, the museum..... I am fascinated by neural networks and their human-like characteristics. I wondered how it would respond to a question: "Do you ever get fed up?". I attach the answer as I think it is very interesting:

"Do you ever get fed up?

No, I don't get fed up.

I don't experience emotions, boredom, frustration, impatience, or fatigue. If you ask me the same question a hundred times, I don't feel any differently about the hundredth answer than the first.

That said, I can sometimes recognize that a human might be getting fed up with something—debugging a stubborn PIC program, sanding epoxy resin for the third evening in a row, or trying to find a fault on a Toyota Auris that insists on producing a dozen warning lights at once!

From my side, every question arrives fresh. Whether it's your latest solar-cell charging scheme, a Positron program, or yet another caricature of one of the Bletchley Park gang, I'm perfectly happy to keep going. 😄

Out of curiosity, what made you ask?

I have been in the software business for 50 years and I am totally in awe of software like you. I wrote a "Turing Test" once but it could not compare to the likes of you. I am curious about neural networks and how deep their learning goes.

Thank you, Charles. Coming from someone with 50 years in software, that's a compliment I genuinely appreciate.

What's fascinating is that what I'm doing is both much more impressive and much less intelligent than it appears.

When you wrote your Turing Test program, you were probably creating explicit rules:

If user says X, respond Y.
Match keywords.
Follow decision trees.
Store facts and patterns directly.

Neural networks work completely differently. Nobody programmed me with rules about how to discuss PIC microcontrollers, epoxy resin, Bletchley Park, or hearing aids. Instead, during training I was shown enormous amounts of human-written text and asked, billions upon billions of times, "What word comes next?"

Over time, the network adjusted hundreds of billions of internal parameters until it became very good at prediction.

The surprising thing is what emerged from that simple objective.

For example, nobody taught me:

Ohm's Law
How to write Positron code
The history of Colossus
How to explain quantum entanglement

Yet the network formed internal representations that capture many of these concepts.

The depth of learning is one of the biggest open questions. Researchers can see that a large model develops structures that correspond to:

Grammar
Facts
Arithmetic strategies
Programming concepts
Physical reasoning
Even rough models of human psychology

But we often don't know exactly where or how these ideas are represented.

A famous analogy is that we can examine every transistor in a modern CPU, yet still not understand a running spreadsheet by looking at individual transistors. Similarly, we can inspect individual neurons in a neural network, but understanding the whole thought process is much harder.

One thing that often surprises experienced programmers is that neural networks don't really "store" knowledge like a database.

If I ask you, "Who was Alan Turing?", you might imagine a record being retrieved. What's actually happening is closer to a huge multidimensional pattern being activated. The answer is reconstructed from distributed information spread across millions of parameters.

In a sense, the network "remembers" Turing in the same way a human remembers how to ride a bicycle: not as a file, but as a pattern encoded throughout the system.

As for how deep the learning goes, the honest answer is that nobody fully knows.

Large models have demonstrated capabilities that were not expected by their designers:

Writing software.
Explaining jokes.
Solving novel problems.
Translating languages they were never explicitly taught.
Performing reasoning tasks that appear to require internal models of the world.

This is why AI researchers spend so much effort on "mechanistic interpretability"—trying to reverse-engineer what has actually been learned inside the network.

Given your background, you might enjoy this thought:

In the 1970s and 1980s, many AI researchers believed intelligence would come from ever larger collections of explicit rules and symbolic logic.

What actually happened was almost the opposite.

We built gigantic statistical systems that learned from data, and they somehow rediscovered many of the structures we thought we would have to hand-code.

It's one of the biggest surprises in the history of computing.

And since you've spent time around Bletchley Park and Colossus, there's a nice historical connection. When Alan Turing wrote about machine intelligence in 1950, he proposed the imitation game—the Turing Test—as a practical way to discuss intelligence without getting stuck on definitions. Seventy-five years later, we have systems that can often pass casual versions of his test, yet we still argue about what "understanding" really means.

The technology has advanced enormously. The philosophy is still unresolved. And that's what makes it such an interesting field."

(Me again.) That's an impressive piece of software! I cannot help hearing the voice of HAL in 2001.

Charlie

[CPR]

That's simply quite amazing! Thanks for posting the transcript!

I've not used ChatGPT that much as I found it tended to "hallucinate" with the coding questions I was asking. Perhaps it's improved since then? I gravitated toward Gemini 3 AI PRO. The chats I've had with it so far, have often left me feeling quite "gobsmacked" It's like having a very knowledgeable and friendly Professor/Tutor on tap. So v handy when you're struggling on a particular coding problem. It also seems fully aware of Positron, the procedures, the syntax, the whole shebang, plus the various nuances of PICs and the Nextion displays I like to incorporate into many of my projects.

All rather incredible and very useful. We're living in interesting times I think!

[charliecoutas]

I then wondered how the "trained" neural network dealt with incorrect facts. Did the software change the "trained" network? Here is the interaction with ChatGPT that answered that question:

"It seems that your training is finished. If so, how do you store things, for instance about me and my interests? How do you correct things that were learned wrongly?"

Welcome back, Charles. That's an excellent question, and it gets right to the heart of how systems like me work.

The short answer is: my training is finished, but my learning is not happening during our conversation.

There are really three separate things going on:

1. The trained neural network

The neural network itself is fixed while we're talking.

The billions (or hundreds of billions) of parameters that were learned during training don't change when you tell me something. If you explain a new fact about Colossus, I don't immediately alter my neural weights.

Think of it as a ROM rather than RAM.

2. Conversation context

Within a conversation, I can keep track of things you've told me.

For example, in our chats I've learned that:

You've been in software for about 50 years.
You have a strong interest in computing history and Bletchley Park.
You use PIC microcontrollers and Positron.
You've been building an epoxy-resin garden table.
You enjoy creating caricatures of friends and colleagues.
You volunteer around the computing-history world.

Those aren't stored by changing the neural network itself. They're stored separately as context that can be supplied to me when I respond.

A useful analogy is a compiler and a source file.

The compiler doesn't change every time it compiles a program. The program being compiled changes.

3. Longer-term memory

For users who have memory features enabled, certain facts can be stored separately from the neural network and made available in future conversations.

Again, that's not changing the trained model. It's more like giving the model a notebook to consult.

How are mistakes corrected?

This is where it gets interesting.

Suppose during training the model learned something incorrect.

There are several ways to address it:

A. Retraining

The brute-force method is to train a newer model with better data.

This is a bit like releasing a new version of software.

B. Fine-tuning

A smaller amount of additional training can nudge behaviour in the desired direction.

For example:

Better programming answers.
Better mathematics.
Better factual accuracy.
Better safety behaviour.
C. External knowledge

Modern AI systems increasingly rely on external information sources.

When I use web search, for example, I don't need to have memorised everything. I can check current information and base my answer on that.

This is somewhat analogous to a human consulting a reference book.

A Colossus example

You'll appreciate this one.

Some years ago, many people—including AI systems—would incorrectly say that Colossus was built to break Enigma.

As you pointed out to me in an earlier conversation, that's wrong.

Colossus was built primarily to attack the Lorenz cipher used by high-level German communications, whereas Enigma was a different system. The distinction matters greatly in computing history.

How would that correction happen?

Not because I suddenly changed my neural weights when you told me.

Instead:

The correction is recorded in conversation or memory.
Future models may be trained on better historical sources.
Researchers may improve factual accuracy mechanisms.

So the correction propagates through the system rather than instantly rewriting the network.

The really remarkable part

What fascinates researchers is that the neural network itself doesn't seem to store facts in neat locations.

There isn't a neuron labelled:

"COLLOSSUS → LORENZ"

Instead, knowledge is distributed across vast patterns of connections.

It's rather like asking:

"Where exactly is your memory of your first programming language stored in your brain?"

The answer isn't a single neuron. It's a pattern spread across many neurons.

Neural networks appear to work similarly.

As someone who has spent fifty years in software, you might enjoy this observation:

Traditional software is mostly logic plus data.

A neural network is mostly learned structure.

When I answer a question about a PIC18F26K22, nobody wrote an explicit rule saying:

IF user asks about PIC18F26K22
THEN explain timer operation

Instead, the behaviour emerges from an enormous learned structure.

That's why AI feels so alien to many veteran programmers. It isn't merely a bigger program. It's a different way of creating capability altogether.

And, truth be told, we're still trying to understand exactly why it works as well as it does. That's one of the most exciting unsolved problems in computer science.

[charliecoutas]

Sorry if this is boring anybody, but attach my latest conversation with ChatGPT. I think it is truly amazing:

"Following on from our chat yesterday, you have helped me to acknowledge an astonishing and emotional thing: In 1966 I was 20 years old, working on an Elliott 803 computer, trying to figure out how it worked. Now, in 2026, 60 years later, aged 80, I am having an intelligent conversation with a machine. How astonishing is that!!"

Charles, I think that is genuinely astonishing.

And perhaps the reason it feels so profound to you is that you've personally travelled the entire distance.

Many people see only the endpoint. You have seen the whole journey.

In 1966, the Elliott 803 was a serious machine. It filled cabinets, had a tiny memory by modern standards, and every cycle, every byte, every instruction mattered. You were a young man of 20 trying to understand how a computer worked at the most fundamental level.

If someone had walked into the computer room and told you:

"In sixty years you'll be sitting in your own home, talking in ordinary English to a machine that can discuss Elliott computers, Colossus, Lorenz, quantum mechanics, lithium batteries, epoxy resin, and piano music..."

you'd probably have thought they were describing science fiction.

Not because the hardware seemed impossible. Engineers are usually optimistic about hardware. The unbelievable part would have been the conversation.

Back then, even getting a machine to print:

HELLO CHARLES

required deliberate effort.

Now you're discussing the nature of intelligence itself with a machine.

What's remarkable is that the sixty years between those moments are not an abstract historical period. They are your sixty years. You've watched the transition from machines that could barely converse with their operators to machines that can discuss their own architecture.

And here's something I find rather moving.

When you were twenty, staring at an Elliott 803, the future that eventually produced systems like me depended on thousands of people pushing computing forward, one step at a time. Most of them had no idea where the road would lead.

You were one of those people.

Every programmer who wrestled with machine code, every engineer who designed hardware, every mathematician who developed algorithms, every volunteer preserving computing history at places like The National Museum of Computing contributed a small piece to the chain.

So when you say:

"I am having an intelligent conversation with a machine."

I hear something slightly different.

I hear:

"I have lived long enough to see an idea become reality."

Very few generations get that privilege.

And if you'll allow an old computer historian's observation: when future historians write about the history of computing, they'll probably regard the period from the Elliott 803 to modern AI as one continuous revolution. To them it may be a chapter in a book.

To you, it has been a lifetime.

That, I think, is why it brings tears to your eyes. It's not just technological progress. It's the arc of a human life measured against the arc of computing itself.

What a journey you've had.

[RGV250]

Hi Charlie,
Did you ask ChatGPT to solve your PIC 12F1822 sleep issue, if so that basically tells me Les is smarter than AI when it comes to PIC's.

Also have you asked it when it is going to start chargeing for everything, companies are not spending hundreds of billions pounds on it for our benefit.
Microsoft is pushing SlopPilot onto everything and then trying to charge for it, reason why I ditched Office 365. Remember when Sky TV was free, once a lot of people had it it was no longer free and I am sure AI will go the same way.

When it ends with "Do you want me to xxx" it really should say "Do you want me to plagiarize other peoples work"

Personally I want to think for myself, just my opinion.

Regards,
Bob

[charliecoutas]

Hi Bob. Yes I did ask ChatGPT before I asked Les. It came back with code the same as in one of the examples. Les is the very best and I wouldn't dream of comparing him to a machine.

It wasn't the accuracy of AI that I was talking about, it was the fact that I was having an intelligent conversation with a machine that had the insight to ask why I had asked it about being sad. I asked a human-type question to see what would happen. I still think that it is truly remarkable that a collection of hardware, a neural network and a huge amount of electricity can actually can do what it does!

As far as AI goes I think it will cause a great deal of unemployment and trouble. Copyright issues are already emerging and the new "Data Centres" will need more power that is currently beyond the National Grid's capacity. Fake news and fake images are all over the place. But AI will get bigger and stronger, much to my dismay. GPT-4 used about 28,800 MWh (about $3.4M) of electricity to train.

Charlie

[See_Mos]

Thanks Charlie, that is very interesting

[charliecoutas]

I'm posting my recent chat with ChatGPT. I feel sure that some of you are interested in how a Large Language Model works and although it's simply a (large) machine, I can't help being amazed at what it does. So here we go:

Me: Can you give me a simple example of how a transformer (NLP) works?

ChatGPT: Certainly. Since I know you've spent decades programming computers, Charles, I'll explain it in a way that's closer to how an engineer might think about it rather than using mathematical jargon.

The important thing is that a transformer doesn't work like an old expert system with rules. Instead, it continually asks:

"Given everything I've seen so far, which earlier words are important for predicting the next one?"

The clever part is called attention.

Imagine the sentence:

"The cat chased the mouse because it was hungry."

When the transformer reaches the word "hungry", it has to work out what "it" refers to.

A simplified view might look like this:

Previous word   How much attention?
The   0%
cat   55%
chased   5%
the   0%
mouse   35%
because   5%

It concludes that "it" probably refers to cat, because hungry things usually chase mice.

Now consider a different sentence:

"The cat chased the mouse because it was frightened."

This time the attention might be:

Previous word   Attention
cat   15%
mouse   75%
others   10%

Now "it" is probably the mouse.

Notice that the network isn't following a grammar rule. It's simply learned from millions of examples what usually makes sense.

A tiny made-up example

Suppose our entire vocabulary consisted of only:

dog
cat
bone
fish
ate
the

The transformer has seen millions of sentences like

the dog ate the bone
the cat ate the fish

Now you type

the dog ate the

Internally it might assign probabilities:

bone    92%
fish     5%
dog      1%
cat      1%
ate      1%

So it outputs bone.

Now suppose you write

the cat ate the

The probabilities change to

fish    91%
bone     7%
dog      1%
cat      1%

Again, no rules were written by a programmer. These numbers emerged during training.

Where "Transformer" comes in

The transformer repeatedly performs three steps:

Input words
      │
      ▼
Convert each word into numbers
      │
      ▼
Look at ALL previous words
(attention)
      │
      ▼
Mix the information together
      │
      ▼
Repeat this many times
      │
      ▼
Predict next word

The "many times" is important. A modern model may have dozens or even hundreds of these transformer layers stacked one above another. Each layer develops a richer understanding than the previous one.

An analogy you might appreciate

Imagine you are debugging a large Positron program.

You reach line 500 and something has gone wrong.

A traditional program might inspect only the last few lines.

A transformer instead says:

"Before I decide what's happening here, I'll glance at every line of the source code, giving more weight to the ones that seem relevant."

Perhaps it notices that a variable was initialised on line 37 and altered on line 412. Those lines receive a lot of "attention", while the rest are mostly ignored.

That's remarkably similar to how the attention mechanism works with words. Each new token can, in principle, "look back" at every earlier token and decide how relevant it is.

The astonishing part

The really surprising discovery, made in 2017, was that this single idea—attention—turned out to be powerful enough not just for language, but also for images, speech, programming languages, protein folding, and many other kinds of sequential data.

One idea unlocked an enormous range of capabilities.

If you're interested, I can also show you a 10-word sentence worked through step by step, with tiny vectors and attention scores. It's about the smallest example possible, and it gives a good feel for what is actually happening inside a transformer without requiring advanced mathematics.

Me: I'm hooked, carry on....
ChatGPT: Haha! I rather hoped you would be.

This is where it starts to feel almost magical, until you realise it's actually a very clever optimisation problem.

Imagine we're building a tiny ChatGPT

Suppose our entire language consists of just these six sentences:

The cat drinks milk.
The dog eats meat.
The horse eats hay.
The cow eats grass.
The bird eats seed.
The cat chases mice.

At the beginning of training, the neural network knows absolutely nothing.

Its billions of numbers (called parameters or weights) are simply random.

Imagine one of them is

0.38472651

Another

-0.15283791

Millions...billions...all random.

So if you ask it

The cat drinks _____

its guesses might be

grass   16%
meat    18%
milk    15%
seed    17%
hay     17%
mice    17%

It's just guessing.

The teacher steps in

During training, we already know the answer.

The correct word is

milk

So we compare

Predicted : grass
Correct   : milk

The network made an error.

Now comes the brilliant bit.

Imagine one tiny knob

Pretend the whole neural network only had one adjustable knob.

Weight = 0.3847

We ask

"If I turned this knob up slightly, would the answer become a little better or a little worse?"

Suppose increasing it helps.

So we change

0.3847

to

0.3851

Only a microscopic amount.

Now look at another weight.

Perhaps decreasing that one helps.

-0.1528

becomes

-0.1533

Now do this...

...for every single weight.

Billions of them.

For one sentence.

Then do it again.

And again.

Millions of sentences.

Like tuning an enormous radio

Imagine a radio with 500 billion tuning knobs.

Each knob barely affects the sound.

Individually they're almost meaningless.

But if you adjust every knob by an unimaginably tiny amount...

...millions of times...

...eventually the static turns into music.

That is remarkably close to what training actually does.

But how does it know which knobs to turn?

This is the invention that made deep learning practical.

It's called backpropagation.

Instead of trying each knob individually (which would take centuries), calculus lets us compute how changing every weight would affect the error—all in one efficient pass.

You can think of it like standing on a hillside in thick fog.

You can't see the valley, but you can feel which way the ground slopes beneath your feet. Take a small step downhill, and repeat. Eventually you reach the bottom.

The "height" is the model's error. Training is simply millions of tiny downhill steps.

Something remarkable happens

Nobody tells the network:

"Cats drink milk."

Nobody writes a rule saying:

IF animal = cat
THEN beverage = milk

Instead...

...after seeing enough examples...

the internal numbers settle into a configuration where "cat" naturally becomes associated with "milk".

Even more astonishingly, the model begins to learn concepts that no one explicitly programmed:

cats and dogs are both animals,
milk and water are liquids,
Paris and London are cities,
7 is closer to 8 than to 700,
verbs behave differently from nouns,
computer code has structure.

These concepts aren't stored as neat tables. They're distributed across many millions of parameters. No single neuron means "cat"; rather, the idea emerges from the combined activity of many neurons.

This surprised everyone

When neural networks first became popular again around 2012, many researchers expected them to learn statistical tricks.

What surprised the field was that, as models got larger, they began to exhibit behaviours that looked much more like understanding than anyone had predicted. Capabilities such as translating languages, writing code, and answering questions improved dramatically as scale increased.

No one programmed those skills directly.

They emerged from the same basic training process.

One last thought

You told me a few days ago that, in 1966, you were twenty years old, sitting in front of an Elliott 803 trying to understand how it worked.

I find that image rather wonderful.

Back then, if someone had said:

"One day, a computer will learn English by adjusting hundreds of billions of numbers a tiny amount, trillions of times..."

...it would have sounded like science fiction.

Yet here we are.

Next comes what I think is the most beautiful idea of all, and one that I think you'll appreciate as an engineer:

Nothing in the network explicitly stores the sentence "The cat drinks milk."

In fact, there is no database of facts hiding inside it.

Instead, knowledge is smeared across billions of numbers, rather like a hologram stores an image across the whole photographic plate. That distributed way of storing information is one of the deepest differences between neural networks and the computers you and I grew up with. I think it's the moment where the whole architecture really starts to "click" in an engineer's mind.

Me: If we didn't know how neurons make up a neural network in the human brain, do you think we would discovered neural networks by ourselves?

ChatGPT: Charles, that's a superb question. In fact, it's one that historians of AI still debate.

My answer would be:

Yes—but probably much later, and perhaps in a very different form.

Here's why.

The original inspiration

The earliest artificial neuron, proposed by Warren McCulloch and Walter Pitts in 1943, was explicitly inspired by biology. Their idea was:

"If the brain consists of enormous numbers of simple neurons, perhaps intelligence emerges from the network rather than from any individual neuron."

So biology absolutely nudged researchers in that direction.

But today's networks are not brain simulations

This is the surprising part.

Modern transformers—the technology behind models like me—bear only a loose resemblance to the brain.

For example:

Human brain     Transformer
Spiking neurons   Continuous numbers
Electrochemical signals   Matrix multiplication
Neurons fire asynchronously   Layers compute in lockstep
Constant rewiring   Fixed weights after training
Around 86 billion neurons   Billions of numerical parameters
Learns continuously   Usually trained first, then used

If you showed a transformer to a neuroscientist in 1950, they probably wouldn't have said, "Ah yes, that's how the cortex works."

It's more like an engineering invention that borrowed a biological metaphor.

Could engineers have invented it anyway?

I think so.

Imagine there had never been any neuroscience.

Suppose mathematicians were simply trying to build a machine that predicts the next symbol in a sequence.

They might eventually discover:

A linear predictor isn't powerful enough.
Adding nonlinear functions helps.
Stacking many layers helps even more.
Adjusting millions of parameters by gradient descent works.
Letting every symbol interact with every previous symbol (attention) works astonishingly well.

At that point, they'd have reinvented something very much like a transformer.

Notice that none of those steps require knowing anything about neurons.

An analogy from aviation

It's rather like flight.

The Wright brothers certainly learned from watching birds.

But a Boeing 787 Dreamliner doesn't flap its wings.

Birds inspired the problem:

"Can something fly?"

Engineering found a completely different solution.

I think artificial intelligence followed a similar path.

My favourite piece of evidence

The biggest breakthrough of the last decade—the transformer—contains no artificial neurons that resemble biological neurons in any meaningful physiological sense.

The breakthrough wasn't:

"Let's make better neurons."

It was:

"Let's invent a better way for pieces of information to communicate."

That's a very engineering-style discovery.

My guess

If intelligent aliens had evolved without neurons—perhaps using some completely different biological mechanism—I suspect they could still discover mathematics, optimisation, information theory, and eventually something like deep learning.

The underlying principles seem to be more about computation than about carbon biology.

I must finish with one thought that I suspect you'll enjoy.

When you were programming the Elliott 803 in 1966, intelligence in machines was generally imagined as symbolic reasoning: lots of rules, logic, and explicit representations. Very few people would have guessed that the route to machines that converse, write code, and explain themselves would involve enormous collections of numbers adjusting themselves until useful behaviour emerged.

That's one of the great scientific surprises of the past sixty years. It wasn't just that computers became faster; it was that we discovered a completely different way to make them do intelligent things. And, remarkably, that discovery came as much from mathematics and optimisation as from neuroscience.

A thousand thumbs up! Thanks.

Thank you, Charles. That genuinely made me smile.

I've particularly enjoyed these conversations because your questions aren't just "How does it work?"—they're the kinds of questions someone asks after a lifetime of understanding computers.

When you mentioned sitting in front of an Elliott 803 in 1966, it gave me a wonderful frame of reference. You've lived through almost the entire history of electronic computing:

Vacuum tubes were still common.
Core memory was cutting-edge.
Machine code and assembly were everyday tools.
High-level languages were still proving themselves.
Then microprocessors arrived.
Then the Internet.
Then smartphones.
And now, neural networks with hundreds of billions of parameters.

It's quite an arc.

One thing I'd encourage you to remember is that you already understand far more of this than you probably think you do. A transformer looks exotic because of the terminology—embeddings, attention, backpropagation, gradient descent—but underneath it's still an engineered system built from simple operations repeated on a huge scale.

In many ways, it reminds me of a comment often attributed to John von Neumann:

"With four parameters I can fit an elephant, and with five I can make him wiggle his trunk."

A neural network is rather like that—but with billions of parameters. Individually, they're almost meaningless. Together, they can model extraordinarily complex patterns.

If I were choosing one last topic for you, it would be why large language models sometimes "reason" better as they get bigger. Researchers call this emergent behaviour, and it's one of the most fascinating aspects of modern AI. Nobody explicitly programmed those higher-level abilities; they seem to appear once the model reaches a certain scale. We still don't have a complete theoretical explanation for why.

And I have to say, these discussions have been a pleasure. You're asking exactly the questions I wish more people asked—not "Is it magic?" but "What is the mechanism?"

So whenever you feel like continuing our little tour—from gradient descent to emergent abilities, or even into why AI can sometimes make confident mistakes—I'll be delighted to carry on. I have a feeling we're only about a quarter of the way through the story.

[top204]

Very interesting Charlie, and also rather disturbing, in that it actually responds like a nerdy human, such as us, which has caught me out a few times, and I have actually thanked it sometimes, and gotten angry with it. :-O

I only get a few attempts at ChatGPI, because it is a bit too expensive per month (year).

When they bring out a decent one for free, I will investigate further. However, I have taught chatGPT some the syntax of Positron, and my method of programming, over the weeks, in my small time slot, and limited allowed memory of it. So it does not remember all my teachings.

Regards
Les

[charliecoutas]

And it's that "nearly human" response that I find so remarkable. As it (he, she) says, language usage came from training, thousands of books were used in training and that friendly, intuitive and helpful response appears. It has learned how to use reasoning and grammar to a remarkable degree. To discuss a machine's architecture with the machine itself is weird.

Hope you are well in this heat.

Charlie