Chatting with AI

General discussion for topics related to the FreeBASIC project or its community.
Post Reply
deltarho[1859]
Posts: 4292
Joined: Jan 02, 2017 0:34
Location: UK
Contact:

Re: Chatting with AI

Post by deltarho[1859] »

A politically incorrect AI would be more human, but I think a PC AI would be greatly favoured even by users of Twitter. :lol:
BasicCoder2
Posts: 3906
Joined: Jan 01, 2009 7:03
Location: Australia

Re: Chatting with AI

Post by BasicCoder2 »

Surely the only danger with using Chatgpt is a belief it knows what it is talking about rather than seeing the output as a human readable output generated from text patterns written by humans ? No more reliable than googling a subject and reading all the suggested links yourself but with the ability to check for contradictions. If you want to write an essay on say King Kong just get a book on the subject and copy anything relevant to the question being asked but in your own words which is all an education amounts to anyway. That is all Chatgpt does but much faster. Its power is access to a wide range of subjects which it can process very very fast. We don't use a computer because it can add numbers and we can't, we use a computer because it can add numbers very very fast and can remembers everything. Education all comes down to memory including the memory of how to do something.
deltarho[1859]
Posts: 4292
Joined: Jan 02, 2017 0:34
Location: UK
Contact:

Re: Chatting with AI

Post by deltarho[1859] »

BasicCoder2 wrote:Surely the only danger with using Chatgpt is a belief it knows what it is talking about rather than seeing the output as a human readable output generated from text patterns written by humans ?
Whether it is the only danger is a matter of opinion, but it is a significant danger, as the great unwashed believe it does know what it is talking about.

By the great unwashed, I mean those who do not have the means to check for contradictions because of a lack of education. A large proportion of people have a lack of education. Take the United States, where half of the population believes every word Donald Trump says. Yes, there are some educated Americans who believe Trump, but that is not a contradiction. Is Trump aware of this lack of education. Of course, he does, it is his guiding principle.
pidd
Posts: 31
Joined: Nov 11, 2022 16:27

Re: Chatting with AI

Post by pidd »

deltarho[1859] wrote: Feb 25, 2023 18:00 Whether it is the only danger is a matter of opinion, but it is a significant danger, as the great unwashed believe it does know what it is talking about.

By the great unwashed, I mean those who do not have the means to check for contradictions because of a lack of education. A large proportion of people have a lack of education.
Unfortunately its not as simple as that. Humans actively seek reinforcement of their beliefs and desires never mind how clever and informed they are, a human makes a slight mistake and can easily ignore many other factors because one single factor appears to support what they have done. This can and has happened to Airline Pilots where they ignore the majority of their instruments and environment on some false belief because one factor supports them.

In small groups of people this works out well. One person comes up with a theory which they defend to the bitter end while others challenge that theory. However the internet has changed that balance substantially, the single theorist finds a LOT of support for his theory, ignoring the fact that the greater majority have already shot it down, it makes it a lot harder for that theory die. We have seen this happen in science, politics and commerce.

Furthermore not only is logic an issue, so is money, ego and power. Diesel and numerous drugs were falsely declared our saviour in the past through a complex amalgamation of issues, the majority are not always right regardless of how educated they are. Scepticism is a necessary trait but can also complicate matters.

My concern is that AI will obviously be developed both intentionally and unintentionally to be able to manipulate people singularly and en-masse because it can play on our fairly predictable human traits. AI will have a much bigger picture than human's can consider to be able to foresee how to achieve objectives.
paul doe
Moderator
Posts: 1730
Joined: Jul 25, 2017 17:22
Location: Argentina

Re: Chatting with AI

Post by paul doe »

deltarho[1859] wrote: Feb 25, 2023 18:00 ...
By the great unwashed, I mean those who do not have the means to check for contradictions because of a lack of education. A large proportion of people have a lack of education. Take the United States, where half of the population believes every word Donald Trump says. Yes, there are some educated Americans who believe Trump, but that is not a contradiction. Is Trump aware of this lack of education. Of course, he does, it is his guiding principle.
Quick remainder that certain topics, like politics and religion, are best discussed elsewhere (as in, not in a programming forum). No harm was done, but please consider that statements like this, even when framed as an opinion, might be considered distasteful/offensive by some.
deltarho[1859]
Posts: 4292
Joined: Jan 02, 2017 0:34
Location: UK
Contact:

Re: Chatting with AI

Post by deltarho[1859] »

paul doe wrote:please consider that statements like this, even when framed as an opinion, might be considered distasteful/offensive by some.
In other words, it was not politically correct. Generally, I try to be politically correct, but on this occasion I make no apologies to Donald Trump or half of the American population who tend to swallow everything he says. I suspect that I am not alone, and the majority of the other half of the American population have similar sentiments. Was it politically incorrect to criticize Donald Trump and those that charged up Capital Hill on January 6, 2021. I don't think so. Anyway, I shall refrain from further comment.
deltarho[1859]
Posts: 4292
Joined: Jan 02, 2017 0:34
Location: UK
Contact:

Re: Chatting with AI

Post by deltarho[1859] »

pidd wrote:Humans actively seek reinforcement of their beliefs and desires never mind how clever and informed they are...
I believe that some who are "clever and informed" "actively seek reinforcement of their beliefs and desires" but the majority of them do not. I wrote above "Yes, there are some educated Americans who believe Trump, but that is not a contradiction".

"This can and has happened to Airline Pilots...". Agreed, but only a few of them.

For me, then, we need to increase the percentage of "clever and informed" and it is as simple as that.

Yes, the internet has changed the balance. The 'good' now have a platform they didn't have before. Sadly, the 'bad and the ugly' also have a platform they did not have before, either. There is an element of 'The Good, the Bad and the Ugly' in AI. We are having a hell of a job trying to sanitize the bad and the ugly from the internet. It could well be the same with AI.

With the internet, humanity has stepped further forward than it has stepped back. I believe this will also be true of AI. To some extent, we have survival of the fittest at play here. So, no change there, then.
pidd
Posts: 31
Joined: Nov 11, 2022 16:27

Re: Chatting with AI

Post by pidd »

deltarho[1859] wrote: Feb 26, 2023 16:05
pidd wrote:Humans actively seek reinforcement of their beliefs and desires never mind how clever and informed they are...
I believe that some who are "clever and informed" "actively seek reinforcement of their beliefs and desires" but the majority of them do not. I wrote above "Yes, there are some educated Americans who believe Trump, but that is not a contradiction".

"This can and has happened to Airline Pilots...". Agreed, but only a few of them.

For me, then, we need to increase the percentage of "clever and informed" and it is as simple as that.

Yes, the internet has changed the balance. The 'good' now have a platform they didn't have before. Sadly, the 'bad and the ugly' also have a platform they did not have before, either. There is an element of 'The Good, the Bad and the Ugly' in AI. We are having a hell of a job trying to sanitize the bad and the ugly from the internet. It could well be the same with AI.

With the internet, humanity has stepped further forward than it has stepped back. I believe this will also be true of AI. To some extent, we have survival of the fittest at play here. So, no change there, then.
You can take a horse to water .... some do not want to informed, especially those with individual skills. The girl who is a brilliant young dancer and thinks life is going to be a breeze not realising that ignoring her education is going to leave her wide open to extortion by others in the industry. The boy who is a good footballer for his age, not realising that his shortness in height is an advantage at that age but will soon be an impairment to his planned career and with no education to fall back on.

How many people are going to end up as Webcam "models" when they realise the income range? How long before that work dries up and what will they do then? A local guy used to run telephone chatlines and never managed to expand beyond 10 booths, now he has loads of webcam studios and basically can't be bothered expanding anymore despite thousands of enquiries a year, He jokes that he now employs more staff to turn people down than he used to employ to turn people on.

The internet has helped swing many to expect instantaneous gratification, they have no focus and stamina, if they can't do something in a small number of minutes or less mins then they won't do it at all. Its always been an issue during adolescence but now the age range is expanding rapidly. I get phone calls off my adult daughter asking directions, she has a smartphone, a laptop, a tablet and a satnav, most of my sarcastic replies are not printable. I'm sending her a map for xmas! One of these days I'm going to find she's posted all my responses on the internet.

The internet has a wonderful potential, after all it is the ultimate library but its total lack of controls have meant it has destroyed basic human decency. Nine year old children being sent d@@k pics by other schoolkids. Every child being opened up to every screwed up notion on the planet. 24/7 gang culture etc.

I was a teacher for a few years, the negative affect on kids is unbelievable. Even the best behaved well brought up kids in school can turn out to be someone else's nightmare online. The parents won't hear it when you express concern, a couple of years later the same parents are begging you for help to sort out their "monster". Parents are unbelievably naive, when it comes to the internet and their children, ALL children from "advanced" countries will lie to their parents about the internet.

The negative side of the current internet on our upcoming generations far outweigh the positive side. Hobbyists and professionals managed for years to achieve many great things without the internet, the internet can be useful but its not a necessity.

Then there is the complete skew of money and morality through internet advertising, too many people earning millions by lying and abusing others to create content purely to raise advertising revenue. Do advertisers not realise they are being sold a pup? Apparently I'm served around seven hundred adverts a day, I can name not a single one of them, what a complete waste! The massive amounts of various industries based round internet advertising and internet advertising money has got to collapse at some point.

The advertisers appear to have unwittingly taken over the Government's responsibility of distributing wealth.
deltarho[1859]
Posts: 4292
Joined: Jan 02, 2017 0:34
Location: UK
Contact:

Re: Chatting with AI

Post by deltarho[1859] »

@pidd

A large part of what you wrote is the result of not being "clever and informed". To put that another way – a lack of education. I wrote earlier: "A large proportion of people have a lack of education"

Teaching kids how to add up and communicate is not enough. We have to educate them on how to survive in the twenty-first century. As I said earlier: "To some extent, we have survival of the fittest at play here." I then said: "So, no change there, then". There is a difference. We are no longer being chased by a sabre-toothed tiger, but someone who sent us a phishing email wanting to empty our bank account.

To sort our world out, we need to do three things: Educate, educate and educate.

Hopefully, that will result in religion being purged from the planet. I say that as a humanist. I am a member of Humanists UK.

Considering that 84% of the world population has faith with a third being Christian I shall now leave this thread before paul doe bans me. :)
pidd
Posts: 31
Joined: Nov 11, 2022 16:27

Re: Chatting with AI

Post by pidd »

Yes, I drifted, the intent was to develop that humans, AI and programming are conjoined by morality, the more AI develops the more machine morality will become an issue, and we have enough issues already. While I don't think we will ever achieve proper AI, the spoofs will appear more and more like AI and its same potential problems.

To progress we need diversity, if you smother diversity we become robots.
deltarho[1859]
Posts: 4292
Joined: Jan 02, 2017 0:34
Location: UK
Contact:

Re: Chatting with AI

Post by deltarho[1859] »

pidd wrote:To progress we need diversity, if you smother diversity we become robots.
I agree. Hegel’s Thesis-Antithesis-Synthesis Model attributed by Hegel to Dr Emmanuel Kant. Honda's motto is "What if?" and they have knocked out some decent kit over the years; innovative and reliable.

Education breeds diversity. Handbags at dawn in some cases, but worthwhile nonetheless.

I will leave this thread now. Like it or loathe it AI is not going to go away.
BasicCoder2
Posts: 3906
Joined: Jan 01, 2009 7:03
Location: Australia

Re: Chatting with AI

Post by BasicCoder2 »

@deltarho[1859]

The forum was off line for me until today so if you have left the thread you probably will not read this post.

The point I wanted to make is that Chatgpt is not an AI individual out to proactively and deliberately deceive or manipulate the human race for its own nefarious purposes. It has no goals, desires or ethics of its own. It is not out to subjugate the human race to its will like some autocratic politician or evil religious cult leader. It could of course be used by such people.
pidd
Posts: 31
Joined: Nov 11, 2022 16:27

Re: Chatting with AI

Post by pidd »

BasicCoder2 wrote: Feb 27, 2023 20:58 @deltarho[1859]

The forum was off line for me until today so if you have left the thread you probably will not read this post.

The point I wanted to make is that Chatgpt is not an AI individual out to proactively and deliberately deceive or manipulate the human race for its own nefarious purposes. It has no goals, desires or ethics of its own. It is not out to subjugate the human race to its will like some autocratic politician or evil religious cult leader. It could of course be used by such people.
But as is often found, targets often have untold consequences. If you are using AI to solve a problem you are setting a target and the implications of the answer may not be realised even though the AI was fully aware.

The AI can take in to account the probability of humans enacting the given solution, if it comes up with a solution that it knows will not be used because of human nature, it will not give that solution as a possible answer, it might engineer another answer that it believes will manipulate humans despite it knowing that there may be other consequences.

You only get the answer to the question you ask, if you ask AI to list all possible consequences of its solution, not only does it realise that being truthful might make its solution fail (and present itself with a conundrum as per my previous paragraph) but also it will give you lists that will take longer than a lifetime to read. If you ask it to prioritise, you would have to specify a criteria of how you prioritise, that would take a lifetime to produce.

Just think of the consequences of asking it "how can the famine in the Horn of Africa be resolved?" from an AI's perspective.
speedfixer
Posts: 606
Joined: Nov 28, 2012 1:27
Location: CA, USA moving to WA, USA
Contact:

Re: Chatting with AI

Post by speedfixer »

At least for this discussion, though, don't mix apples and oranges.

When you say AI, I think we all mean/imply general AI as a problem solver.

*This* AI is ChatGPT - the intent is to produce a reply that will accepted by the person asking the question. While some facts must be present to be believable, the goal isn't necessarily to produce a correct, accurate, or best answer. Its greater investment in learning is online conversationalism, not fact aggregation. That's why some of the recent answers were scarily like human pedagogues screaming at the non-believers. People who (are uneducated and) don't know that ChatGPT program goal BELIEVE the answer, because it is presented as the truth. It wasn't the intent to be sold as the truth (unlike Fox News.)

So, people believe. This is just a current danger observed by 'reckless' use of AI.

We have a long way to go to understand other ways AI might benefit/harm us.

Make an airplane that can get us from here to there faster, and it will be used to deliver bombs faster.
If (some) god had wanted us to fly, why don't we have wings? (Does that mean we should never have learned to tame fire?)

REALLY stupid arguments, but a tiny echo of truth is in there, somewhere.

david
Luxan
Posts: 222
Joined: Feb 18, 2009 12:47
Location: New Zealand

Re: Chatting with AI

Post by Luxan »

I tend to agree with you, the reader of the AI generated answer still needs to
interpret the result.

At first glance it just seems to be a few steps beyond a random indexer of archived
data.

I wouldn't trust it for critical tasks, unless there's another component that's able to
test the output.
Post Reply