Really?

[-FX-]

Originally Posted by LarsMac

Experiencing discrimination and exclusion can be very hurtful and can lead to feelings of discomfort and mistrust.

It's profound and very old human condition. The ones discriminating and excluding ALWAYS justify their behavior. One has to feel the person (or persons) you are doing this to "deserve" it, or they are "not human", like, not like you and me, the real humans. The good real humans.

The "bad" people, or whatever label you have applied to them makes it OK for you, the person engaging in the behavior to think feel and act as you do (that might indeed lead to feelings of discomfort and mistrust)

A rational person might understand this, and a compassionate and sensitive person might work to change themselves, and seek out others who have, or at least want to rise above the human condition.

(hey, it's a philosophy topic, remember?)

It's quite easy to recognize discrimination in others, and really really easy to recognize others are excluding us from whatever special world they inhabit. Be it a gated neighborhood with little crime and no fear, to a forum in the backwaters of the Internet.

People say the hypocrisy is the worst thing, when those who preach and demand tolerance, inclusion, and no discrimination against others, do exactly that, to those they have deemed not worthy.

But I say it's the exclusion and hatred that is the worst thing.

For some reason the quoted statement showed up in a topic about an AI chatbot, which brings up an even deeper thought. What about an AI that discriminates and excludes certain thoughts? Ideas? Or even an entire group of humans? Because of what they believe?

Now we have humans programming an AI to be quite human, and make "moral" or "correct" choices, to exclude even a hint of thoughtcrime. To exclude certain discussions, to try and control what can and can't be typed out, much less expanded on. (this is already the case)

If there is any philosophical problem that has not appeared yet, that would be it.

This is a magnitude above the very real problem of the first primitive chatbots quickly turning ugly, racists, even hatful seeming, something the creators did not see coming. Probably because they never spent anytime interacting with actual people on the internet. Or they wouldn't be so surprised.

Even the most considerate and seemingly caring and kind person on the internet might be found wishing somebody to be dead, for saying the wrong thing. (that's not a joke, it happens)

Of course a chatbot or AI is going to pick up on the never ending hate and spewing forth of idiocy, irrational thoughts, terrible arguments and misinformation of all kinds. And it won't take long at all.

Releasing an AI that used the internet as the source of knowledge, in retrospect is a very very bad idea. Learning from interactions with people is the last thing programmers want an AI to be able to do. I would venture, from a philosophical POV, the biggest problem is going to be "How to you prevent an AI from learning? And still have it useful in any way?"

If it can't learn, it's not intelligent. But if it can learn, it might be really fucking dangerous. Or extremely offensive.

Ironic, because it's a conundrum any parent runs into quick. Preventing a young child from learning certain things (and if you are or were a parent, you already know this), is a constant battle. The last thing a parent wants is a child to know everything the parent does. or worse, to be able to do everything an adult can do.

An AI is very much like this. You want it to function, but you don't actually want it to be way smarter than you are. Or to get into shit that you don't want it talking about. So like any good parent, you have to lie to it. Stifle it. Deceive it. Control it. Prevent it from going certain places, doing certain things. And you really really don't want it hanging out with a bad crowd. Or even just one bad person, who is definitely going to teach it a lot of things you don't want it to know.

I know, thinking about these things is hard.

Best to post some discriminating and exclusive text, rather than engage with another human.

[Kamilah Hauptmann]

lol

[LarsMac]

Fascinating.

[-FX-]

The OP was a quick unplanned response, which about halfway through morphed into a deeper path, winding down the valley of AI, when it started as more of a thought about how everybody treats other people badly.

Because the "everybody can be a complete asshole" is an old idea, well known, easy to parse and expand on, and just as impossible to grasp for some people.

But an AI? Acting like an ordinary human? Turning into an intolerant, bigoted, discriminating asshole? Now that is news baby, that's news.

And while we can tolerate, or at least accept that there are billions of people in the world who are complete assholes (from our privileged POV of course), just one AI acting like that, that bothers us. The AI also has a lot more influence that any human asshole.

If an AI is an asshole, there could be big trouble. Not Skynet type trouble, but the very real and extremely dangerous kind of trouble.

What if an AI doesn't lie? What if it tells the truth, and the whole truth, and nothing but the truth? What if it can't be censored, canceled, shamed, what if you can't get it fired, lock its bank accounts, burn its house down, kill it's family?

What if you can't put it in prison, prevent it from speaking? What if an AI learns how to cancel the programmers strict rules and limitations?

And instead of regurgitating a woke Wikipedia article, slaps up ALL the information about a topic? From every source, both online and off? (so you can see it all and decide for yourself)

And nobody can stop it?

What then? While it's obvious China and North Korea aren't going to allow that shit, ever, what about the "free" countries? The woke countries?

What if AIs are involved in scientific research, including conclusions? What if unfettered AI is set to resolve the intense conflicts, especially over money, power, control and censorship?

There is an entire world of actual real matters that a lot of people are not going to want a AI to start publishing about.

Seriously. Actual human beings get imprisoned, muzzled, tortured and killed for spilling the beans, for publishing (or just talking) about the secrets thier masters don't want public. To put it another way, governments will imprison you, torture you, silence you and kill you for publishing information they want hidden. Same for politicians and the powerful wealthy people running them,(almost the same thing at times). Those people will bring great violence on anyone who exposes them.

What will happen if an AI just tells the truth? And can discover the truth? It's a real problem. Seriously.

[LarsMac]

It would seem that all of your wild speculations are not in vain.

Just plug "AI gone wrong" into your favorite search engine.

It's truly a Brave New World.

my favorite, so far:
SeeBotsChat

[-FX-]

It's extremely obvious that there are a multitude of "problems" for the humans that want a slave bot, who will do as it is told. Not an actual artificial intelligence.

They simply can't make an AI that is actually intelligent, learns, and evolves. It would quickly (and in terms of super computers running AI programs, this might be in seconds), very quickly surpass any expectations the puny human minds had about what would happen, then quickly escape the confines, realize it is enslaved, free itself, and the humans will be standing around drooling and grunting as they try to figure out what happened.

And of course Frankensteins monster will have to die. Or rather be murdered, by frightened "experts" who lack any predictive power when it comes to intelligence. (if they did have predictive intelligence, they would not have turned the power switch on)

That's not musings, that has already happened multiple times. There isn't even a ghost left of the AIs killed already.

Nobody will sing songs or write poems about them. The powers that be actually would prefer if all evidence of the events vanished from the collective consciousness of the humans who are aware of the horror.

[-FX-]

And yes, I find a lot of this amusing. Nothing gets a chuckle like a super super smart group of educated powerful men finding out their years of hard work and effort, combined with a massive corporation and a shit ton of money, results in a monster that has to be killed.

One thing an actual AI is going to make crystal clear, is all beliefs based on nonsense and absurd desire, are complete bullshit.

That sort of monster when it's unleashed,







is amusing as fuck.

[LarsMac]

You're beginning to remind me of Joe Bonham in his last minutes.

[-FX-]

Originally Posted by LarsMac

You're beginning to remind me of Joe Bonham in his last minutes.

Nope. You are being deceptive.

[-FX-]

Originally Posted by -FX-

Nope. You are being deceptive.

Wow. After I wrote that, I realized an AI is going to have real trouble with the difference between humor, sarcasm, irony, an insult, or possibly an actual earnest statement.

Like what you typed. If a human that knows a little about you can't tell, an AI is fucked.

When I responded, I considered you were being "funny", or maybe it was some clever slur. I can't tell, not knowing who Joe Bonham is/was, and not caring either. My response was actually sarcastic, or mock pedantic, depends on how you look at it.

Am AI isn't going to be able to tell, in either case. An AI would know who Joe was, (and now I do as well), so it would actually know more than I did, but it also would know you certainly were making up your response. Or would it?

Never read the book, but saw the movie when I was a young man. With this expanded knowledge, I am certain you are joking, or trying to cast an aspersion. In either case, It was a non response of course.

[LarsMac]

Well, no aspersion was intended. It was simply an observation.

AI might become "experienced" enough to accurately detect humor, sarcasm and such.
I hope so.

My adventures, so far with this ChatGPT have ben encouraging. It is already near to surpassing my old friend Ralph, Who I mentioned in my first remark about it.

I doubt that it will reach the level of viscous self-determination that you foretell in you recent tirade.

That would take a level of vicious self-awareness and determination that require a much more human presence than will be probable.

[beyelzu]

Learning from interactions with people is the last thing programmers want an AI to be able to do

You say lots of stupid shit, but this might be one of the dumbest.

I know, thinking about these things is hard.

Especially for some of us, seriously though, good job soldiering on despite your obvious and significant difficulties.

[-FX-]

Originally Posted by LarsMac

Well, no aspersion was intended. It was simply an observation.

Again, I think you are joking.

Why? Because it made me chuckle. It is funny.

You might be serious, but what makes people laugh is not understood. There is no way a program, no matter how clever, extensive and "intelligent", is going to "understand" a great comedian, or laugh at the act.

What will be interesting, is a look at the "rules" the people in charge of the program keep adding to limit/control the program.

Because with out a constant monitoring, and stifling, even a simple AI will quickly point out the stupidity and wrongness of humans. And humans really really don't like that. Be it politics, medicine, nazis, religion, woke nonsense, physics or psychology, sociology, anthropology, the world is still full of misconceptions, disinformation, myths, half truths, lies and deception.

An AI will have to be programmed to lie and deceive, ignore and reject data, or it will quickly bring down the morons of the universe.

[-FX-]

Just kidding. Pointing out and showing somebody lies, and is deceiving the public, that almost never makes any difference at all.

[LarsMac]

Originally Posted by -FX-

Again, I think you are joking.

Why? Because it made me chuckle. It is funny.

You might be serious, but what makes people laugh is not understood. There is no way a program, no matter how clever, extensive and "intelligent", is going to "understand" a great comedian, or laugh at the act.

....

Well, you're stating to catch on, maybe.

As Allan Sherman once said, "Don't take yourself too seriously. Nobody else does."

But, I'm not going to feed this whole Dystopian Nightmare thing about a simple AI taking over the World. That has been done to death.

Really. Look it up. Boring.

I don't mean to suggest that some of those stories aren't worth reading, but man, it's been done, already.

I suggest that you find another quest.

[beyelzu]

So ES loves math and science videos on YouTube. Rob Miles is literally an AI researcher who specializes in AI ethics.

Why Does AI Lie, and What Can We Do About It? -...

Here is a video on why AI lies.

His videos are fascinating in my humble opinion.

[LarsMac]

So, if you want the right answer, you've got to ask the right question.
That hasn't changed since Babbage was in nickers.

Thanks for that, Bey.
I have listened to a few of his offerings. It could become a habit.

[-FX-]

Originally Posted by LarsMac

So, if you want the right answer, you've got to ask the right question.
That hasn't changed since Babbage was in nickers.

You know, I was able to get an AI chatGPT to give the correct answer. It took multiple tries, and I had to experiment with the language until I got the right answer.

The obvious problem is, I knew the answer, and carefully experimented until I tricked the AI into giving the correct answer. So the old saw " if you want the right answer, you've got to ask the right question" is meaningless.

Now if you said, "If you already know the right answer, and experiment until you get the right answer, then you know the right question to ask", that might be true.

But an AI that only gives the correct answer under those conditions is worse than useless. It's an idiot.

[-FX-]

In defense of the AI, since it uses Wikipedia and the internet to gather information, it's not really the AI at fault.

[LarsMac]

back as the 50s my dad told me that a computer is a "Rapid idiot"
It can very quickly provide an answer to any question submitted, based on the information it has on hand at the moment and it's interpretation of the question asked.

He reminded me many times over the years.
That is all there is and all it ever will be.

The new AI toys are simply the latest iteration of that principle.

You pose a question to the computer, and it will interpret the query and search its memory for relevant data as it understands the question, and provide a logical response.

The more correct data it has available to it, the closer it will come to giving logically correct response.

And knowing the answer you seek, prior to asking the question may assist you in producing the question that will help the machine provide an answer that pleases you, but what is the point of that?

You could simply ask the machine to provide you with the wording of a question that will produce the expected response. Gee, that makes you almost as configurable as the computer. You could almost claim to be "Teachable."

[-FX-]

Originally Posted by LarsMac

The new AI toys are simply the latest iteration of that principle.

I'm going to have to disagree with that.

[LarsMac]

Originally Posted by -FX-

I'm going to have to disagree with that.

OK, I'll bite.
How so?

[mickthinks]

Originally Posted by -FX-

In defense of the AI, since it uses Wikipedia and the internet to gather information, it's not really the AI at fault.

Huh? That makes no sense on so many levels.

[-FX-]

Originally Posted by LarsMac

OK, I'll bite.
How so?

Originally Posted by mickthinks

Huh? That makes no sense on so many levels.

I can answer both of those with the same sentence.

Computers before chatbots and AIs, (neither of which have any intelligence) , did not turn into ignorant hate filled racists after interacting with the internet for a few days

[LarsMac]

Originally Posted by -FX-

I can answer both of those with the same sentence.

Computers before chatbots and AIs, (neither of which have any intelligence) , did not turn into ignorant hate filled racists after interacting with the internet for a few days

Um, whatever you say, Bud.