While many mistake Expert-Systems for AI...

it’s all in programming, when you say a program that can rewrite itself while running, define rewrite.

your question of the difference between ‘opinion or evaluate’, I believe all opinions come from an evaluations. Human or AI. With AI, it’s based on conditional parameters, yes, ot can be, as more base line data is collected. An example is below.

My brother put in a robotic milker, to milk cows a few weeks ago. The ‘robot’ monitor each cow, and a number of different parameters.

Each cow has a RFID tag, and the cows are tracked through out the day (laying down, going for water, going for feed, going to grooming as well as milk production, this is drilled down further to quarters (teats) of the bag with location and quality, it records temperature, analysis of the milk, the condition of the teat locations, bag size, cow size, number of times the cow volunteered to be milked, when the cows in ‘heat’ to name a few.

Any red flags that falls outside the perimeters sent, a text would be sent immediately to my nephew and hired herdsman. And it tracks trends also and makes ‘programmed’ adjustments.

I would consider this AI, because a set of parameters are created after a base line is set for each cow. The more data collected, the more accurate adjustments are made.

I won’t call this Re-writing a programs, but I would call it, a self-adjustment a program makes through subroutines.

right now, I look at AI as a tool, that is taking away, responsibility from a human intervention on mundane tasks, and doing a better job of it.

and to a degree, it can recognize where it needs a human to intervene, just like a human, and alerts the human with a message to people that can help, where by the way, some humans have a hard time with Asking for help.

that’s today, I don’t know what tomorrow will be.

1. There is a definite, separate, time bounding, between the writing and running phases: as a metaphor, "you cannot fire while you cock the hammer, and you cannot cock while you fire"

- Even in a chain-reaction of a machine gun, such separate time bounds are alternating, by cocking, then firing, cocking, firing etc., meaning the compiled program cannot "write itself" (? - whatever that may mean) while running, and the compiled program cannot run while writing itself, whatever "writing itself" may mean (hence the picture of Baron Munchausen, lifting himself and the horse he rides from the bog they are drowning in).

But hey, that's only the tip of the iceberg:

2. And even if such fantasy could happen, according to which self-conclusion would the program re-write itself while running ?

- How can a running program generate a conclusion derived (processed) by recent precedant-input. Any such input would either halt the program idle in an endless loop (no such input was anticipated by the programmer) or cause the program deny processing (ignore) the precedent, exempting itself by generating and error message.

But that's just my own opinion.

Nevertheless, the O.P meant to ask : "How would you define Artificial Intelligence", and to be more particular, how it would differ from any Big-Data Expert-System, such as, say, Deep-Blue, or, say, Auto-Piloting a jet or a car, or, say, Stock Market "buy or dump advisory system", or, say, Chat-GPT - the list can go on.

I don’t know what you mean by fantasy,… 7 less you mean this thread itself.

on determining self-conclusion, I agree, you can not cock the hammer as you fire, you or the programs,… needs to evaluate ‘data sets’ that is collected as time goes on and compare them to the program conditionals that has been programmed.

as far as defining A.I, that’s a subjective matter based on each individual, that has no definite answer.

https://www.nytimes.com/2021/09/09/technology/codex-artificial-intelligence-coding.html

"An expert system is a computer program that uses artificial intelligence (AI) technologies to simulate the judgment and behavior of a human or an organization that has expertise and experience in a particular field. Expert systems are usually intended to complement, not replace, human experts."

"In artificial intelligence, an expert system is a computer system emulating the decision-making ability of a human expert. Expert systems are designed to solve complex problems by reasoning through bodies of knowledge, represented mainly as if–then rules rather than through conventional procedural code. "

It’s interesting what you was said, about Re-writing a program as the program while running.

I forgot all about this, very early in my correct, I tried to create a programs, where variables that would redefine itself as the program runs. I called it a dynamic variable…I felt I would come close, but always had a syntax errors…

What I was trying to do is create a program that would eliminate the human element… with minimum amount of variables… but storing values in a multi-dimensional type of array…

I spent 3-10 hours a day after work trying to make it work, I felt I was close, but that was due to inexperience…

"A program which can re-write itself while running"

There actually is such a thing, a Neural Network.

The program consists of connection weights between neurons connected in adjacent layers. Some are programmed by giving it an input and the right answer, but some learn autonomously, categorizing input data.

I find Neural Networks fascinating, but I'm not ready to let one drive my car...

Something good about artificial intelligence, is that in attempting to build such a device, that intelligence must first be defined...The vetting of data by determining validity, relevant nature, speed of change, legal and moral appropriateness, social repercussions of ratification and other criteria can set the groundwork for better education methods and efficiency .. in other words, we must first know the true nature of intelligence, to reproduce it....

The confidence level must also be determined, low confidence level requires more information....theory testing, if this is so, then this must be so...

Absolutely, and just imagine; suppose there is a program which re-writes itself while running: any false trivial fact or a bit of data which is fed into it's data crunching system is a potential bug - as Humans we have the natural ability to re-evaluate our whole set of assumptions once we discover we were fed with BS or false info

'There actually is such a thing' can be also said about Cold Fusion, Artificial Intelligence, Machine-Learning, etc.: It's more of an ideal or goal, than it is an actual, concrete reality.

I asked a friend once: Do you believe God really exists? His reply was: 'Anything you may think of, really exists. It exists in your mind'...

As an example : the actual fact, that there is a whole research field dedicated to the development of Cold Fusion, does not suggest there is such an actual technology up and and running, or that such technology is even remotely feasible

The concept of robots building other robots is akin to reproduction in the biological domain.

One of the five concepts that a living thing possesses is respiration - to use its own biological systems to generate its energy needs. Robots don't do that, though a solar-charging robot, as a concept, would come awfully close to the same behaviour as that of a plant. If it were able to build other robots, does that count as "being alive" (further definition required)?

Dearest Slack, good to meet you back after so long.

I suspect you made a quantum leap raising this discussion to realm of "I Robot" - which is an exciting discussion by itself (nuclear energy power source?), however, my intention in posting this discussion was much more humble: 'How would you define Artificial Intelligence'

- Would you expect Machine-Learning (AKA "Deep-Learning") be similar in nature to that of humans?

- What's the difference (in computer terms) between 'Smart' and 'Wise'?

That sort of thing...

Would this be consider AI or an expert system.

Expert-System - Professionally-oriented Big-Data based real-time manager.

AI is what the O.P's trying to define.

How would you define AI ?

What do you imagine it being able to do ?

I would define it as levels of AI, depending on the goals it’s being used for,

as far as defining one level to the next. Never gave that much thought.

Having said that, can you distinguish between an interactive app, and an active information system which has its own identity, motivation, and self-inspired goals?

Subjectively, I’d say yes,