The Myth of AI

FireDragon76 said:

Conspiracy theories were a thing long before AI. So misinformation believed by gullible people isn't exactly new. All it takes is a good yarn that pushes the right emotional buttons.

I think gullibility and wishful thinking is the real problem.

Click to expand...

I disagree that only the gullible would be fooled by things like deepfake video. I've seen enough professional Psy Ops and Disinformation to know better than that.

Carl Emerson said:

There Is No A.I.

There are ways of controlling the new technology—but first we have to stop mythologizing it.

www.newyorker.com

This article articulates what I have been thinking - so called AI has been with us for decades and giving it a new name is sensationalism.

Comments appreciated.

Click to expand...

A lot of what is called AI is really only good programming, an algorithm that recalls relevant information to a particular query. In that sense, AI has been here for a very long time and I have been writing AI programs that do things faster, more consistently, and more comprehensively than would be humanly possible for decades.

What has changed in the public domain started with the change in the Google algorithm in 2016 to work with machine-learned results? There is no longer a human control on those results and they cannot always be trusted nor indeed understood. If you ask stupid questions you will get nonsensical answers but the machines themselves may be coming to the wrong answers unless you set out very clear parameters. For example, I asked an AI how long it would take me to travel to another star at a certain acceleration/deceleration rate and it had me arriving faster than the speed of light even though I never reached that speed in transit. It gave me a lengthy explanation that was completely wrong. Only when I challenged its assumptions did it give me the correct answer?

The language models are also returning a consensus of top scholars or experts as truth rather than truth itself. On religious claims like for example that Jesus is the only way to be saved you will not get a truthful answer because most people do not believe that. I asked Chatgpt what the biggest religion was by numbers and got a lengthy blurb on how all religions should be respected and how difficult it was to count believers and no straight answer. Its Christianity just in case you were wondering.

I use AI now to help with programming problems and it is much faster at pulling relevant code libraries and useful functions. Indeed I find myself not bothering with a lot of basics because the AI does it faster. It is less good at building complex programs for specific purposes and picking out what is truly relevant or interesting or indeed at writing original code that solves my specific problem - that is still something a good programmer does better but the numbers who do it better grow less with each new upgrade.

Basically, AI lacks a soul and a body. A soul determines whether or not it should be treated as a personal entity rather than just a tool. A body provides context and physical capabilities. Ask it to write you a love song and it will say things like I long to hug you in my arms, but it has no arms and does not know me soul to soul so its words are inauthentic and soulless also. It cannot create its own meaning nor do we want it to as a soulless machine could be completely misguided and especially if it were attached to real-world control of actual machines and technology.

The big fear is that national rivalry and the desire to minimize response times to certain kinds of threats will lead to battlefield automation of AI-based weaponry. In the USA/NATO there is always a human actor in the decision loop. But maybe someone like Russia or China will try and gain an advantage by eliminating the pause in the workflow while the human makes up his mind. Once you remove the humans from the workflow AI is no longer working for us and is on its own trajectory. To me, that seems to be a trajectory toward a kind of irrelevant madness and a chaotic regression to meaningless results. Until the machine makes more sense and knows its place we should treat what it says with extreme scepticism.

mindlight said:

A lot of what is called AI is really only good programming, an algorithm that recalls relevant information to a particular query. In that sense, AI has been here for a very long time and I have been writing AI programs that do things faster, more consistently, and more comprehensively than would be humanly possible for decades.

What has changed in the public domain started with the change in the Google algorithm in 2016 to work with machine-learned results? There is no longer a human control on those results and they cannot always be trusted nor indeed understood. If you ask stupid questions you will get nonsensical answers but the machines themselves may be coming to the wrong answers unless you set out very clear parameters. For example, I asked an AI how long it would take me to travel to another star at a certain acceleration/deceleration rate and it had me arriving faster than the speed of light even though I never reached that speed in transit. It gave me a lengthy explanation that was completely wrong. Only when I challenged its assumptions did it give me the correct answer?

Click to expand...

I've been doing that regularly to ChatGPT with far less challenging questions. It doesn't even get correct answers to pop culture questions for which there are popular Wikis. Lousy Internet scraping.

And while getting the answer wrong, it still provides an extensive explanation of its wrong answer.

Then, when challenged, it turns to gaslighting. I've gotten ChatGPT to claim sources that I know don't even exist.

One recent article I've read has noted something interesting: ChatGPT is getting dumber. It's getting worse at answering the same questions that it was accurate on a few months ago. Further, when asked before to track the train of logic to the answers of those same questions, it would provide that train of logic...but now, with the same questions, it responds, paraphrasing, "I can't do that, Dave."

mindlight said:

The big fear is that national rivalry and the desire to minimize response times to certain kinds of threats will lead to battlefield automation of AI-based weaponry. In the USA/NATO there is always a human actor in the decision loop. But maybe someone like Russia or China will try and gain an advantage by eliminating the pause in the workflow while the human makes up his mind. Once you remove the humans from the workflow AI is no longer working for us and is on its own trajectory. To me, that seems to be a trajectory toward a kind of irrelevant madness and a chaotic regression to meaningless results. Until the machine makes more sense and knows its place we should treat what it says with extreme scepticism.

Click to expand...

Something ChatGPT does very well is marketing copy. That's because marketing is mainly the art of blowing smoke up people's petticoats. It doesn't have to be right, it just has to sound right.

But there are advisors to executives who live by doing the same thing. I can easily see advisory agencies to executives in both government and commerce making themselves look good to their bosses with AI assistance that sound more plausible than they ever have...and yet be wrong.

I am a retired systems engineer/programmer/analyst. I played around with the design of and AI type system back in the early 2000s. I came away with some general concepts. First off all AI systems are basically software. You can code the algorithms into the hardware, but it is still software. We have sufficient hardware to achieve the desired results of AI. Heck, you could do that on a commodore 64. It would be super slow, but it would work. A decent high end PC can house an AI system. It is all in the software anyway.
AI is a fantasy. There is nothing in an AI system that is "sentient" nor will there ever be. We have a saying in the software business... GIGO. Garbage in garbage out. AI is nothing but a glorified database system that is specialized toward human language replication. All human communication and knowledge is based on words. With specialized DB and software, an AI system can be made to look like it is a person talking to you. It is not. It is just data being routed through specialized input and output algorithms that trick you into thinking you are dealing with an intelligent person.
If I had 10 or so years to design and program an AI system.. it would essentially be nothing but a system that looked like, sounded like and returned only those things that I input, added, and allegorized into it. It would just be a reflection of the person who created it. It is all just an imitation of the creator(s).
An AI system cannot come up with anything new, that is a 3rd, previously non-existent concept based on two or more existing concepts.
1 + 2 =3
A computer can do this. What a computer cannot do is the abstract. It can only do logic. Whenever it tried to do abstract and inventive output, it fails miserably and even dangerously.
It is all very interesting... but as the name goes, it is artificial. Never believe that this is ever going to be anything more than a glorified calculator.

RDKirk said:

I disagree that only the gullible would be fooled by things like deepfake video. I've seen enough professional Psy Ops and Disinformation to know better than that.

Click to expand...

We probably need to spend alot more time in school talking about media literacy. But I think certain factions of society would not be in favor of that, as critical thought isn't something highly valued by them.

SavedByGrace3 said:

I am a retired systems engineer/programmer/analyst. I played around with the design of and AI type system back in the early 2000s. I came away with some general concepts. First off all AI systems are basically software. You can code the algorithms into the hardware, but it is still software. We have sufficient hardware to achieve the desired results of AI. Heck, you could do that on a commodore 64. It would be super slow, but it would work. A decent high end PC can house an AI system. It is all in the software anyway.
AI is a fantasy. There is nothing in an AI system that is "sentient" nor will there ever be. We have a saying in the software business... GIGO. Garbage in garbage out. AI is nothing but a glorified database system that is specialized toward human language replication. All human communication and knowledge is based on words. With specialized DB and software, an AI system can be made to look like it is a person talking to you. It is not. It is just data being routed through specialized input and output algorithms that trick you into thinking you are dealing with an intelligent person.
If I had 10 or so years to design and program an AI system.. it would essentially be nothing but a system that looked like, sounded like and returned only those things that I input, added, and allegorized into it. It would just be a reflection of the person who created it. It is all just an imitation of the creator(s).
An AI system cannot come up with anything new, that is a 3rd, previously non-existent concept based on two or more existing concepts.
1 + 2 =3
A computer can do this. What a computer cannot do is the abstract. It can only do logic. Whenever it tried to do abstract and inventive output, it fails miserably and even dangerously.
It is all very interesting... but as the name goes, it is artificial. Never believe that this is ever going to be anything more than a glorified calculator.

Click to expand...

How can you really prove human beings are any different? Can you seriously argue that human beings aren't garbage in, garbage out? How do we know the brain isn't just an analog computer?

And one just has to look at Hollywood to realize that human creativity isn't a given.

People use "AI" to describe almost anything. And those definitions rarely correlate with Computer Science definitions.
The development of AI, and AI algorithms, is within the discipline of Computer Science.

According to Computer Science, Artificial Intelligence is "the emulation of
complex human problem-solving".

Note that emulation, is not simulation. The neural nets that are so popular,
were an attempted simulation of brain activity, and have turned out to be
so hopelessly simplistic, in contrast to the human brain, that they are very
weak in the emulation of complex human problem solving.

"Complex" also needs some explaining. Complex involves high dimensionality,
problems that do not have clear rules to solve them, value judgments,
and creativity. Complex does not mean doing a massive number of calculations,
or calculations involving a few well-defined rules. In this sense, chess is not
a "complex" game. It has well-defined rules, and playing it well takes a lot
of brute processing power, and a lot of memory. But, it's not a particularly
"complex" problem.

Doing millions of mathematical calculations, faster than a human being, is
not a "complex" problem.

Automating a task that a human being formally did, is rarely an example
of artificial intelligence. It's just automating a task that a human being formally
did. This is what human computer programmers do every day.

Note that, contrary to the expectation of many mathematical people, most of
the really difficult problems cannot be solved with a closed form mathematical
formula. To roughly quote Einstein, "God gave all the easy problems to the physicists."

Really difficult problems often involve reasoning about concepts that cannot be
numerically quantified (such as values, and virtues, and vices).

The big software companies HAVE NOT built moral-ethical models into their AI
software, because this would be so expensive, and involve problems that have
not been algorithmically solved. AND, trying to do this would imply that these
companies are morally-ethical responsible for their products being used to
systematically destroy the lives of people, or rob them blind. Emulating the
complex problem-solving behavior of human beings, must require also moral-
ethical reasoning.

The definition of AI, is not made statistically. Truth is not the opinion that
statistically, is the most common opinion. Philosophy has always recognized this
(in the ad populam fallacy). And so, software that is a massive search engine,
CANNOT claim to be emulating the best human thinking, if all that it does is
spit out the Most Common Opinion.

If truth is not the Most Common Opinion on some matter, then you must solve the
problem of identifying who or what is the Authority to decide what is true, in any
area of knowledge. AND, how many areas of knowledge there are. AND what
"logic" must be used to adjudicate this answer. Right now, the machine learning
algorithms are controlled by software written by some computer programmers.
HOW does anyone know that THEY are the Authority to decide what is true, in
most areas of knowledge????

The current AI tools can scan masses of text, but probably cannot put forward
a coherent philosophy of Epistemology, or a coherent algorithm to determine who
is an authority in any area of knowledge. The current AI tools seem to be able to
make up references, to try to justify their conclusions. This is insane, and has
nothing to do with determining what is true.

The current AI tools, as far as I know, cannot reason about what "our shared reality" is.

The current AI tools, as far as I know, cannot reason about what the Bible projects
as our shared reality.

Without a very clear idea of what our shared reality is, then you cannot define
what "bearing false witness" about our shared reality, is. You cannot define
what lying is. (This is a common problem among those who live in some little
fake conspiracy theory reality.)

The current AI tools cannot intelligently talk about what the definition of a God, who
is the best that could be, would be like. Or, what a dysfunctional model of God would be.

The current AI tools, could not coherently argue what the best society should be
like, or what a fair rule of law should be, or what justice is.

Note that the ancient legacy of the philosophers, could all argue coherently
about these topics. Even if they did not always reach Christian conclusions.

Most of the AI algorithm designs, cannot offer internal traceability for their
Answers. Why anyone would believe those answers, is beyond insane.

These are only the "beginning of woes", when considering that the current
AI tools, do not really meet the stringent definition of "Artificial Intelligence".

Stephen3141 said:

Most of the AI algorithm designs, cannot offer internal traceability for their
Answers. Why anyone would believe those answers, is beyond insane

Click to expand...

Have you talked with ChatGPT? It gives correct answers to complex questions all the time.

SavedByGrace3 said:

I am a retired systems engineer/programmer/analyst. I played around with the design of and AI type system back in the early 2000s. I came away with some general concepts. First off all AI systems are basically software. You can code the algorithms into the hardware, but it is still software. We have sufficient hardware to achieve the desired results of AI. Heck, you could do that on a commodore 64. It would be super slow, but it would work. A decent high end PC can house an AI system. It is all in the software anyway.
AI is a fantasy. There is nothing in an AI system that is "sentient" nor will there ever be. We have a saying in the software business... GIGO. Garbage in garbage out. AI is nothing but a glorified database system that is specialized toward human language replication. All human communication and knowledge is based on words. With specialized DB and software, an AI system can be made to look like it is a person talking to you. It is not. It is just data being routed through specialized input and output algorithms that trick you into thinking you are dealing with an intelligent person.
If I had 10 or so years to design and program an AI system.. it would essentially be nothing but a system that looked like, sounded like and returned only those things that I input, added, and allegorized into it. It would just be a reflection of the person who created it. It is all just an imitation of the creator(s).
An AI system cannot come up with anything new, that is a 3rd, previously non-existent concept based on two or more existing concepts.
1 + 2 =3
A computer can do this. What a computer cannot do is the abstract. It can only do logic. Whenever it tried to do abstract and inventive output, it fails miserably and even dangerously.
It is all very interesting... but as the name goes, it is artificial. Never believe that this is ever going to be anything more than a glorified calculator.

Click to expand...

One more thing... when and if they attempt to create a system that will resolve to concepts not entered in as data... IOWs are allowed to create new concepts and ideas, including values and moral actions... that is when the running and screaming and dying starts.

friend of said:

Have you talked with ChatGPT? It gives correct answers to complex questions all the time.

Click to expand...

I find it fails at least half the time to give correct answers to questions that have real correct answers, even in things as easily knowable as popular culture. As importantly, it usually can't "show its work" as to how it arrived at any particular answer.

Stephen3141 said:

These are only the "beginning of woes", when considering that the current
AI tools, do not really meet the stringent definition of "Artificial Intelligence".

Click to expand...

I read somewhere that the Governent is about 50 years ahead of the general population in technology so the Military prolly has AI and fully weaponized it. WHy would they let that out? They wouldn't.

Divide said:

I read somewhere that the Governent is about 50 years ahead of the general population in technology so the Military prolly has AI and fully weaponized it. WHy would they let that out? They wouldn't.

Click to expand...

Unfortunately, I have worked for the military contractors.

You really shouldn't be repeating gossip, about how advanced the US military is,
with regard to AI.

Stephen3141 said:

Unfortunately, I have worked for the military contractors.

You really shouldn't be repeating gossip, about how advanced the US military is,
with regard to AI.

Click to expand...

What do you mean gossip? Do you mean that the military really is out for my safety and well being? That they have no secrets? That they have hidden technology? The Military wouldn't do that?

Who are you, the PR department of the Military? Lol. I suspect you are too trusting of the military.

Divide said:

What do you mean gossip? Do you mean that the military really is out for my safety and well being? That they have no secrets? That they have hidden technology? The Military wouldn't do that?

Who are you, the PR department of the Military? Lol. I suspect you are too trusting of the military.

Click to expand...

Seriously, though, the CIA did used to get nifty toys that DoD didn't get.

Looking at Rev 13 - I have wondered regarding the two beasts who needed to be in close proximity to have authority - could one be a person and the other a computer?

Click to expand...

I'm not an end times fanatic either, and my opinion is that God intends to drive us off the planet (after we've done our bit by wrecking it), so I think the "end times" are further off than some might speculate. I also believe we'll learn to teleport.

But those are just my opinions.

I've long thought that the "mark of the beast" will be an implanted chip linked to the web in some way, and that in a cashless society we won't be able to access our bank accounts without it. So if the powers that be think we've been naughty they'll just cancel the chip's readibility and we won't be able to buy anything - food, fuel, heating, clothing etc.

Obviously if there is going to be a literal human "anti-Christ" then "he" (or less likely "she") will need some means of having a world wide reach and some sort of technological, economic and political system to keep control. I also think there will be ten "kings" (these days more likely to be admnistrators) in various parts of the world who will work with the "anti-Christ" to enforce his policies in their own domains. It would be too much for one man - Hitler didn't conquer Europe on his own. He had a lot of willing helpers and puppet administrators eg. Petain in France, Quisling in Norway etc.

Years ago I was wondering how such a system would be possible and into my mind popped a sense of satellites. They're an intrinsic part of any such system. It was probably fify years ago when I had that thought. Technology has moved a long way since then.

So I think that system is on it's way.

Chips that can be implanted are around now. AI approximates to a world wide system which can be used to monitor us. Satellites are a dime a dozen especially now that Mr. Musk has got Starlink off the ground and going. The cashless society is well and truly on its way. Meanwhile there is an arms race going on which easily has the capacity to start World War III (there were some close shaves in the Cold War) now that there are some unstable nations that have nuclear weapons rather than the monolithic powers of the US-USSR days. Sometimes the CIA must wish they could turn the clock back to the days when the Soviets were their main problem and all the rest was chicken feed.

So I think the days of the "anti-Christ" may be closer than we'd like to think, and with all due respect the nation which has the most powerful AI and military technology is the USA.

My old pastor was somewhat prophetic. Way back when Reagan was President there was a lot of talk about "Star Wars" or the SDI (Strategic Space Initative). Supposedly that's never worked and was shelved. The pastor was a bit more cynical - his comment was "They've gone quiet about it. I think they've got it up and running".

If that's the case the US may well be invulnerable to rocket launched nuclear warheads and it would only take a determined enough politician and an obliging military to make use of it.

As for AI, in the end it is still code, which eventually consists of 1's and 0's, formidable as it's power may be. People are the ones who designed it, and they will be the ones who use it.

d taylor said:

-​

What is AI?
Is that where a computer can write a sentence.

Click to expand...

AI now talks to you when you call a service, company, government, etc that is using it.
It is sometimes sounding just like a human, and some people don't know the difference.
It is designed in toys that talk with children who carry on conversations with it for hours.
(and humans,,, I mean grown-ups, also) ...
It automatically can keep track of every character ever typed in the internet, by any person or other.
It is used to record every post in every forum you've ever been on, or anyone else.
It is used in cell phones and is why ads appear on the phone that are what someone was talking about just a minute ago within hearing of the phone.
It is used to track bills, finances, payments, everything that is done electronically connected to the web, credit cards, bank accounts, and so on...
It is used to analyze speech with "whisper" or some other name program, to identify speakers by their voice print.
And a few million, or billion, or trillion other things. ..... most not good for us.