|
"Reinforced Learning with Feedback" is the algorithm that ChatGPT is based on.
With a huge dataset taken from the internet and from the feedback of everyone that uses it telling it correct or incorrect it will "learn" but it won't learn the same way humans do.
It won't have imagination.It won't have intuition. It won't have any of the human characteristics that make what we humans call Intelligence.
That's why it'll always be ARTIFICIAL INTELLIGENCE. An IMMITATION of intelligence. Never the real thing. Let's get real. It's no more than just a system running an algorithm. If we treat is as such and it's no more than just a fancy toy. Treat it like your new "GOD" and well "It'll become faster, smarter" and all of that lah-dee-dah. Time to choose, humans.
|
|
|
|
|
If it's not linked, it's "unknowable". There will be a new movement to not record anything; we'll just exchange information with winks and nods so it can't be used.
"Before entering on an understanding, I have meditated for a long time, and have foreseen what might happen. It is not genius which reveals to me suddenly, secretly, what I have to say or to do in a circumstance unexpected by other people; it is reflection, it is meditation." - Napoleon I
|
|
|
|
|
So the ubiquitous cameras will be used by the AI to learn the "winks and nods" language.
Resistance is futile!
Freedom is the freedom to say that two plus two make four. If that is granted, all else follows.
-- 6079 Smith W.
|
|
|
|
|
Not being an anthropologist or a sociologist or whatever, I have recently been thinking (because I can) about the nature of "intelligence" (and sapience).
Particularly when people assert that many of the other animals on our planet are also sapient and intelligent -- which I don't deny.
But that means that "something else" must separate humans (and probably our extinct proto-human ancestors) from "the lower animals" -- but not "intelligence".
Because certainly "something" separates us from even our closest opposable-thumbed relatives.
Either that or there is disagreement on what constitutes intelligence -- I'm probably too stupid to have it explained to me.
While we humans have amassed a large collection of knowledge and technical ability I don't think we have actually become more intelligent than our stone-age forebears. In particular, we have the technology to transfer knowledge to others.
Consider: One day a cave man -- who did not have a stone hammer -- made a stone hammer, the first stone hammer ever conceived in the history of mankind. He did not become more intelligent by making the stone hammer, he already had the intelligence required to do it. But he gained knowledge and technical skill. He could then teach others to do the same, and those others gained knowledge and technical skill, but not intelligence -- they also already had the required intelligence to grasp the concept and the benefits to their lifestyles.
Regardless what the CIA thinks, I don't think knowledge and skill equate to intelligence.
In short, I don't think an AI has intelligence at all, it has only knowledge -- and logic programmed into it about how best to manipulate and present that knowledge.
The only way to win is not to play the game.
|
|
|
|
|
PIEBALDconsult wrote: But that means that "something else" must separate humans
Does it? Is it really a categorical difference? Could it be so simple as elephants not ending up with opposable thumbs, for example? Or rats didn't lose their fur so that they could develop oral communication, etc?
I'd argue that if there is a thing that separates us from animals, it's probably a combination of things. Opposable thumbs, highly articulated language, the inclination to alter our environment rather than adapt to it. All of these things led us to the top of the proverbial pecking order in nature's hierarchy of life.
I don't know if there's any one thing that's fundamentally different than other animals, at least in terms of category. Other animals have language (dophins), but not as articulate. Other animals alter their environment (beavers), other primates have opposable digits, as do many birds.
With the disclaimer that I am not religious: I remember reading the fall of man story in the old testament and coming away thinking that (based on my exegesis) part of what it was saying is that what makes us different than the rest of the animals is the ability to define and articulate elaborate moral frameworks ("the tree of knowledge of good and evil"). I thought that was interesting.
To err is human. Fortune favors the monsters.
|
|
|
|
|
So I think we are basically in agreement. 
By "something else" I don't necessarily mean it has to be one thing, it may be a subtle variety of smaller things combined.
Yes, there are "lower animals" which do some of the things we do, such as communicating, using simple tools, making shelters, altering the environment (e.g. beavers making ponds). But we're the only ones who do all of them and more and to a mind-boggling extent. Surely our ancestors learned from them.
I don't think we can make assumptions about the morals of other animals, especially cats.
|
|
|
|
|
PIEBALDconsult wrote: But that means that "something else" must separate humans (and probably our extinct proto-human ancestors) from "the lower animals" -- but not "intelligence".
The simplest concept of what that "something else" is, is the ability to choose. Animals generally respond instinctively (they can be trained to not respond instinctively, but that's still not choice.) We humans are unique in that we can choose not to respond by instinct. Ooh, that cake looks delicious, I'm going to eat it. Or, nice cake, but I'm watching my calories. Or I'm lactose intolerant so eating that would not be a good idea. Conversely, my cat loves to chew on certain plant leaves regardless of how many times he barfs them up later.
Therefore, I would say that intelligence is making good choices based on knowledge and skill, and also making poor choices for reasons we are conscious of but choose to ignore.
|
|
|
|
|
I'll need to give that more thought. Personally, I'm unclear on what constitutes instinct anyway, so I may be a bit lost.
As to choice, I'd still be unsure where to draw the line. For instance: When a pack of predators attacks the weakest members of a herd of prey, is that instinct or choice? Wouldn't instinct demand they attack the largest/meatiest? Is attacking the weakest members a learned strategy?
This reminds me of "A Beautiful Mind".
I think humans have probably lost much of the instinct our ancestors must have had and replaced it with learned knowledge.
Maybe that's what makes the difference today, but there still must have been chooser-zero who had the ability and acted on it. Probably some bratty kid refusing to eat his mammoth.
|
|
|
|
|
PIEBALDconsult wrote: Is attacking the weakest members a learned strategy?
The most amount of energy for the least amount of work, plus some morale/long-term thinking? Human babies could grow up and eventually help us so we let them be, some animals think very differently. (This sort of went dark awfully fast). At any rate: AI could definitely do that since its basically just math:
Calculus of variations - Wikipedia[^]
|
|
|
|
|
PIEBALDconsult wrote: Is attacking the weakest members a learned strategy?
Certainly is learned for larger carnivore mammals. Packs are easiest to see this but even when non-pack animals younger animals often have to survive on smaller prey because they keep picking the wrong prey animal to attack.
|
|
|
|
|
As your cat does with the leaves.
|
|
|
|
|
nepdev wrote: A school should prepare kids for life, give them some competence they can use, some knowledge they can apply, make them curious to create and use their imagination.
This is why I sent my kid to a Waldorf School K-12, fortunately had the ability to do so. I was stunned when a couple months ago he called me simply to say how he so appreciated that I had done that. That was nice. 
|
|
|
|
|
I've been saying this ever since I was in elementary school.
The school system is inherently flawed.
Curiosity is punished rather than awarded.
It does some things good, given that most of us are at least a somewhat functional adult (stretching the definition of functional a bit there  ).
But recent study, for example, showed that kids in the Netherlands stop reading when they go to school.
It has to do with the mandatory reading.
Apparently, you're allowed to read LOTR or Harry Potter, but all schools talk about is rather ancient reading because that's what the teachers know.
The books that are spoken about in detail today are the same as the books in my time, which were the same books my parents had to read.
Once you've read a few of those book you'll learn to hate reading.
I may not be the best example since I've always hated school, but I loved programming.
Until I went to school for it.
During that time I haven't touched Visual Studio in my spare time for months.
When I quit it took a while for me to get active again.
Eventually, I got some of my enthusiasm back.
All in all, kids are getting worse in almost every subject, especially language and math.
|
|
|
|
|
Logical Song: When I was young, it seemed that life was so wonderful
A miracle, oh it was beautiful, magical
And all the birds in the trees, well they'd be singing so happily
Oh joyfully, playfully watching me
But then they send me away to teach me how to be sensible
Logical, oh responsible, practical
And they showed me a world where I could be so dependable
Oh clinical, oh intellectual, cynical
There are times when all the world's asleep
The questions run too deep
For such a simple man
Won't you please, please tell me what we've learned
I know it sounds absurd
Please tell me who I am
I said, watch what you say or they'll be calling you a radical
Liberal, oh fanatical, criminal
Won't you sign up your name, we'd like to feel you're acceptable
Respectable, oh presentable, a vegetable
Oh, take it take it yeah
But at night, when all the world's asleep
The questions run so deep
For such a simple man
Won't you please tell me what we've learned
I know it sounds absurd
Please tell me who I am, who I am, who I am, who I am
'Cause I was feeling so logical
D-d-digital
One, two, three, five
Oh, oh, oh, oh
It's getting unbelievable
Listen here: The Logical Song ( Lyrics ) Supertramp - YouTube[^]
modified 16-Feb-23 12:11pm.
|
|
|
|
|
nepdev wrote: Is schooling really meant to be repeating random facts, regurgitate what you have been told so you can spit it out again on an exam paper? Is this "learning"?
For the most part the answer is yes. Despite what many might claim I see little evidence that formal education in its entirety teaches people to 'think'. Might get lucky in some cases but for the most part it is just a matter of learning the exact date of '1492', where the comma goes in the sentence and how to add and subtract.
nepdev wrote: Cramming data down their throat is, in my opinion, NOT what a school should do.
Many others have that opinion as well. And there have been many attempts to provide an alternative way. But none of those work (they seem to work in very, very small tests but in larger rollouts they change nothing or even lead to more problems.)
So certainly if you know a way that can provide revolutionary change to education then you should step up and start proving it and then popularizing it.
nepdev wrote: Even Einstein apparently said "most of my work came from imagination, not logical thinking. And if you have problem with mathematics, I assure you mine are still greater."
The educational system targets the average not the above average. That is necessary because, by definition, most of the users (students) of the system are and always will be average.
nepdev wrote: Something needs to change here, doesn't it?
Discard the calculator? Discard the keyboard? Discard the slide rule? Discard pencil and paper? Should students be taught solely by making new copies of religious texts?
Or recognize those are changes that do in fact help students.
If AI agents were capable of, in general, answering all questions correctly then why not use such a useful tool. Doing so does not directly lead to no one thinking - it is just another tool.
However nothing suggests that AI agents are even close to being capable of providing correct answers to even most things. They provide answers to many things but the validity is often in doubt and is in fact often wrong. But in the same way if you rely solely on the results of a calculator to build a bridge you should expect that it will fall down (probably even before it starts being used.)
|
|
|
|
|
jschell wrote: Many others have that opinion as well. And there have been many attempts to provide an alternative way. But none of those work (they seem to work in very, very small tests but in larger rollouts they change nothing or even lead to more problems.)
Hear! hear!
If I may add a metaphor. A juggler who can keep many balls up in the air needs two things: the dexterity and the balls. Without the balls, only the dexterity will not help. During school years our brains acquire the ability to connect disparate facts in new and surprising ways but they need those facts. Knowing that they exist on Google or Wikipedia does not help. They need to be inside the brain so they can be rotated, rearranged, looked at from a different vantage point for in the end, with a bit of luck, the brain to come up with that magical thing, a novel idea.
The theory of education pendulum has moved from rotten memorization to creative thinking but there is a golden midpoint where you teach young brains both facts and how to operate with those facts.
Mircea
|
|
|
|
|
After having taught my children basic math, algebra, geometry and trig - I absolutely despise walking away from "rote memorization". The add/sub/mult/div tables dealt with the fundamentals of basic facts that got children over the details that actually allowed them to think. They get to Algebra - which is a fascinating time - and rather than wrestle with basic math, they can focus on the abstract concepts.
Creative thinking as it were... that golden mid point. But Bureaucracy gets paid for elephanting basic concepts to prove they need a job.
Charlie Gilley
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
Has never been more appropriate.
|
|
|
|
|
charlieg wrote: I absolutely despise walking away from "rote memorization"
Not to mention how do you teach them discipline, drive, focus, working towards a goal and completing it?
Certainly seems to me that taking a test once a week to regurgitate what was covered in the previous week (or month) has a chance of producing some positive improvement in those that I mentioned.
charlieg wrote: Creative thinking as it were
Never ever seen any alternative studies that could demonstrate that creativity was actually being taught. But certainly can measure whether a 10 year knows how to add two numbers together.
|
|
|
|
|
I dunno. Take out loans. Go to grad school. Have debts forgiven. Live long and prosper.
"Before entering on an understanding, I have meditated for a long time, and have foreseen what might happen. It is not genius which reveals to me suddenly, secretly, what I have to say or to do in a circumstance unexpected by other people; it is reflection, it is meditation." - Napoleon I
|
|
|
|
|
nope. student loans are a way to slavery. Hoping they get forgiven is just stupid.
Charlie Gilley
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
Has never been more appropriate.
|
|
|
|
|
|
Me? I'm cashing out my home equity to invest in HVAC and septic. Everyone wants to be warm, cool and their toilets work.
Charlie Gilley
“They who can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety.” BF, 1759
Has never been more appropriate.
|
|
|
|
|
charlieg wrote: invest in HVAC and septic.
Go one level lower, and become a plumber. Those HVAC and septic companies need skilled workers!
Freedom is the freedom to say that two plus two make four. If that is granted, all else follows.
-- 6079 Smith W.
|
|
|
|
|
In my teaching days (community college level), I created the final exams. Each problem was structured according to the Bloom taxonomy of learning. The students could bring any printed material to the exam - the course textbook or anything else.
The 'a)' question asks for 'simple facts' that can usually be copied directly from the course textbook, 'knowledge'.
'b)' is for the student to show that (s)he understands the meaning of facts, 'comprehension'.
'c)' asks the student to demonstrate how the understanding of facts is used to solve a specific task, 'application'.
'd)' asks for an explanation of a suitable breakdown of a complex situation/system, 'analysis'.
'e)' asks the student to combine various elements / principles / ... into a larger, more complex whole to create a solution, 'synthesis'.
'f)' asks for a critical, 'professional' evaluation of some technique / solution / ..., 'assessment'.
In recent years, many people put 'synthesis' (aka. 'creating') at the very top of the learning pyramid, 'assessment' (aka. 'evaluation') at the second level from the top. I tend to agree more with the original ordering: I have met lots of people - both in programming and in other fields - sprouting creative ideas like a fountain, producing lots of results, yet completely unable to do any sort of critical evaluation / assessment of either their own creations or the works of others. The other way around: In order to make a true assessment, you cannot be a stranger to the process of synthesizing a whole from constituents; you must master it quite well. Assessment is going a step further in mastering your field.
It did take some practice & experience to create exams suitable to tell how far up the pyramid a student could climb, but after a handful of exams, as I got the grip on how to phrase the a) - f), I could quite easily see where the candidate was starting to fall off, not mastering that kind of questioning. Or, the candidate was a true master in the area presented in 1a) to 1f), but didn't handle problem 2) much higher than to 2d), clearly weaker in synthesis and assessment in that area.
I am quite sure that ChatGPT would handle most problems up to the c) level quite well - although I guess I could fool it by 'trick questions' that would be handled by a human. For d) to f), it is not that difficult to phrase the problem statement so that a mechanical, robotic search would easily be revealed as a fake.
If I were creating exams nowadays, I would of course test them with ChatGPT in advance, both to learn even better to make 'trick questions' and to learn how lack of true understanding can be disguised. For the higher learning levels, I suspect that ChatGPT would reveal itself by being far more wordy, talkative, than a person with a true understanding that would go straight to the point. At least that is my impressions of ChatGPT dialogs I have seen. (I have only watched others using ChatGPT; I haven't chatted with it myself. I did have some chats with Eliza way back in 1975, though.)
|
|
|
|
|
The toughest exams I did at university were "open book" exams where we could take any printed materials (books) in with us. The exams tested our understanding of the material and ability to use it in novel ways. In other words it tested our mastery of knowledge, not the knowledge itself, i.e. our ability to think.
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
