Artificial Intelligence: ChatGPT

[tomh207]

@ChrisInOz indeed, the invention of what would become the most prevalent database in the world, excel.

 

Tom

[Knebo]

Hi all, 

This is a bit delicate and I don't want to be offensive to any new members... but I've noticed a couple of new members joining over the last few hours, that may well be AI creations or phishing attempts. All young ladies...

Just be careful if you receive personal messages asking you for your credit card details 

Cheers

[tomh207]

  On 2/6/2025 at 3:18 PM, Knebo said:

Hi all, 

This is a bit delicate and I don't want to be offensive to any new members... but I've noticed a couple of new members joining over the last few hours, that may well be AI creations or phishing attempts. All young ladies...

Just be careful if you receive personal messages asking you for your credit card details 

Cheers

Expand  

Thanks for alerting us. We have had the usual cycle of bots this week. All new members are moderated for a period and we keep an eye on them and deal with them appropriately if required.

 

Tom

Edited February 6 by tomh207

[Neverenoughwatches]

  On 2/6/2025 at 3:18 PM, Knebo said:

Hi all, 

This is a bit delicate and I don't want to be offensive to any new members... but I've noticed a couple of new members joining over the last few hours, that may well be AI creations or phishing attempts. All young ladies...

Just be careful if you receive personal messages asking you for your credit card details 

Cheers

Expand  

Saw that 3 in fact in 3 hours.

  On 2/6/2025 at 4:47 PM, tomh207 said:

Probably not posted yet, we should take this back of house I think Richard.

 

Tom

Expand  

Tom , just have a look at the profiles, one has way too much information on.

[tomh207]

The users in question have been investigated and removed from the forums. If anyone has anything to report like this please use the report function or message the moderators.

 

Thanks

 

Tom

[MikeEll]

  On 2/5/2025 at 1:18 PM, ChrisInOz said:

AI doesn't "understand" anything at all,  ever.   Understanding is not a feature of its design.   It is a sophisticated pattern matching algorithm,  that strives to match phrases from a question with whatever it has managed to dredge up from the festering cesspit that is most of the Internet.

Since there are some sites, like this one,  on the internet, that have genuine information,  AI will sometimes fluke onto a sensible answer,  but don't bet your house on it.

Really,  referring to it as "intelligence" should simply be regarded as false advertising.

 

Expand  

well, the keyword that you've dropped is 'artificial'   It's a programming technique that aims to emulate intelligence, and to some extent, it does achieve that.  Answers are improved over time as more information is learned from training.  I get what you mean, though.  

What I find interesting is that the information obtained from actual 'intelligent' humans can be equally erroneous.  Anyone who's studied for a degree will know that all information needs to be evaluated, regardless of where it's obtained.  It can be correct, accurate, inaccurate, biased, misleading, or, in some cases, plain wrong.  I think the point here is always to educate yourself and reach your own opinion rather than blindly trust any source of information.

[AndyHull]

  On 2/6/2025 at 1:38 PM, ChrisInOz said:

The name of this miraculous product?  FORTRAN.

Expand  

There is  a certain irony in the fact that knowledge of Fortran is still in use in some niche areas (mainly banking and finance) to this day. Maybe I'll ask Chatgpt to write me some Fortran 77 code, just for fun.

Speaking of AI and time (wasting), you might find this interesting.

https://hackaday.com/2025/02/06/a-great-use-for-ai-wasting-scammers-time/

 

[tomh207]

  On 2/6/2025 at 7:53 PM, AndyHull said:

There is  a certain irony in the fact that knowledge of Fortran is still in use in some niche areas (mainly banking and finance) to this day. Maybe I'll ask Chatgpt to write me some Fortran 77 code, just for fun.

Speaking of AI and time (wasting), you might find this interesting.

https://hackaday.com/2025/02/06/a-great-use-for-ai-wasting-scammers-time/

 

Expand  

COBOL as well, know a couple of folks who work 4-6 months of the year and earn 6 figures working with it.

 

Tom

[RichardHarris123]

We know AI doesn't understand anything but we don't know if or when it will do. 3 months, 3 years, 30 years, 300 years or never?

[AndyHull]

  On 2/6/2025 at 8:19 PM, RichardHarris123 said:

We know AI doesn't understand anything but we don't know if or when it will do. 3 months, 3 years, 30 years, 300 years or never?

Expand  

As currently implemented, possibly never.

The problem is that we don't actually fully understand how organic life forms "understand", so it is difficult to define understanding in the context of anything else, including AI, however current AI does not work the way most people think it does, and certainly does not work the same way the human brain for example works.  By that definition, it will never understand.

That is not to say that we will never be able to make a system that is truly self aware, and  genuinely understands its own existence. AI is however, not that system.
 

[JohnR725]

  On 2/6/2025 at 7:53 PM, AndyHull said:

There is  a certain irony in the fact that knowledge of Fortran is still in use in some niche areas (mainly banking and finance) to this day. Maybe I'll ask Chatgpt to write me some Fortran 77 code, just for fun.

Expand  

While you're at it why don't you check and see If chatGPT can write a cobalt program for you. 

[AndyHull]

  On 2/7/2025 at 6:09 AM, JohnR725 said:

While you're at it why don't you check and see If chatGPT can write a cobalt program for you. 

Expand  

I'll tell you what, I'll see if it can write me a Cobol compiler ... entirely in Fortran 77.

  Quote

A full COBOL compiler in Fortran 77 would be a massive undertaking, but I can certainly write a very simple COBOL interpreter or a minimal subset compiler that translates a small subset of COBOL into an intermediate representation or directly into some Fortran code. Would you like a basic interpreter for COBOL statements, a translator to Fortran, or just a parser that checks syntax?

Expand  

Well it looks like ChatGPT would be up for the challenge.

[JohnR725]

  On 2/7/2025 at 4:11 PM, AndyHull said:

Well it looks like ChatGPT would be up for the challenge.

Expand  

thinking about how wonderful this is I wonder if it could convert a clipper program in DOS to something much newer that runs in Windows? Then I'm curious as to how many people reading this know what clipper is with out looking it up? In other words no cheating and looking at Wikipedia.

 

[tomh207]

  On 2/7/2025 at 9:06 PM, JohnR725 said:

thinking about how wonderful this is I wonder if it could convert a clipper program in DOS to something much newer that runs in Windows? Then I'm curious as to how many people reading this know what clipper is with out looking it up? In other words no cheating and looking at Wikipedia.

 

Expand  

Clipper database?

very vague recollections about software called clipper back in the DOS, C/PM days of yore

 

Tom

[ChrisInOz]

  On 2/6/2025 at 7:53 PM, AndyHull said:

There is  a certain irony in the fact that knowledge of Fortran is still in use in some niche areas (mainly banking and finance) to this day.

Expand  

Yes,  I had a nice contract back in the noughties,  converting a Fortran system with flat files into Java with an Oracle database.  Fixed a few ancient bugs while I was at it.

  On 2/7/2025 at 4:11 PM, AndyHull said:

I'll tell you what, I'll see if it can write me a Cobol compiler ... entirely in Fortran 77.

Expand  

Funny you should suggest that.  One of my projects for the final year of my degree,  back in 1979,  was an Algol compiler written entirely in Pascal.   Never used either of those languages in my career later.  In the eighties,  almost everything was COBOL.

[VWatchie]

  On 11/18/2023 at 11:43 PM, nickelsilver said:

Chatgpt, ha, it sounds better already than most of the YouTubers.

Expand  

ChatGPT can't yet generate video content, but it can create illustrations. The above image illustrates "curb pins". Enjoy!

I especially like the screw slots. I haven't got the screwdrivers for them yet but I guess it's just a matter of time before Bergon has them on offer.

[JohnR725]

  On 2/7/2025 at 9:38 PM, tomh207 said:

database

Expand  

in amusement that I have is database programs are very important in the universe but seemed to be almost invisible in the programming world.

  On 2/7/2025 at 9:38 PM, tomh207 said:

very vague recollections about software called clipper back in the DOS, C/PM days of yore

Expand  

so according to Wikipedia company started in 1984. sold the company in the early 90s where the program continued for several more years then did not transition into Windows. Although part of the problem was that the programmers programming clipper in DOS had a hard time transitioning into Windows also.

  On 2/8/2025 at 8:41 AM, VWatchie said:

ChatGPT can't yet generate video content, but it can create illustrations.

Expand  

there are programs out there that generate videos. you're seeing more and more examples of that of people making videos for YouTube. I wonder if we can find somebody with the video generation program and ask it to make a running watch for us?

Then the illustration is interesting of what happens when? I don't suppose you could ask to produce a slightly different version of illustration in other words I like to pull it apart so we can see all the bits and pieces and how this new watch works?

[AndyHull]

  On 2/8/2025 at 8:41 AM, VWatchie said:

especially like the screw slots. I haven't got the screwdrivers for them yet but I guess it's just a matter of time before Bergon has them on offer.

Expand  

I'm sure I've seen those screws before. I think they were in a Mumbia special that had been recently "serviced". They too were  bit of a multiple choice question when it came to how to undo them.

[VWatchie]

  On 2/6/2025 at 1:38 PM, ChrisInOz said:

Most amusing of all was the day I discovered that I had been "redundant" before I was born.   A technical article from 1958 breathlessly announced that all computer programmers were now obsolete,  since an amazing new product would allow the end user to simply tell a computer what they wanted,  and it would all be done for them at once.

The name of this miraculous product?  FORTRAN. 

Expand  

I remember back in 1987 when I first started to learn programming. A friend of mine told me not to put too much time and effort into it as something called AI (Artificial Intelligence) soon, within a year or two would be doing all the programming and I'd be out of work. I've been making a living as a programmer and teaching programming ever since.

Now we're clearly in the hype era of AI. I'm old enough to have experienced many hypes and they never delivered what they promised, but I guess this time will be the exception, or!?

The "hypes" change our lives, but never as predicted and as fast as expected.

Edited February 10 by VWatchie

[Knebo]

  On 2/10/2025 at 12:03 PM, VWatchie said:

within a year or two would be doing all the programming and I'd be out of work.

Expand  

[AndyHull]

I'm in two minds about AI's abilities to write code. It tends, currently to write code that is the programming equivalent of that bizarre "curb pins" watch drawing. What it boils down to is the inescapable fact that AI has no ability to comprehend the real world. It is debatable whether it ever will.

Don't get me wrong, ChatGPT and its pals write code. They can write it in pretty much any language, and they can write it quickly. However they need a lot of prompting to make the code actually functional. In that respect they are both good and bad at the same time.

If you need a framework written quickly for a well defined problem, then AI can, to some extent, do that.

You still need to go through the iterative process of refining the code to make it actually work. In other words the AI doesn't (yet?) have the ability to do this, so it is still down to the porridge between your ears to get things to function, and probably always will be.

Will I be out of a job any day soon, as a result of advance in AI? Probably not.

Will over reliance on AI be  the cause of problems in the future? Probably.

Will AI, on balance, be a benefit or a hindrance to us?

I suggest AI is a tool. It is not the only tool in the toolbox, and it is a relatively narrow use tool. Whether it is a net benefit to the human race, all rather depends on what we choose to do with it. It isn't going to do anything we humans don't ask it to do, so if we use it intelligently then it will benefit us.

Given the state of the world today, I would say the chances of us using it intelligently are probably about 50-50. 

In other words, don't give up the day job just yet. Robbie the robot is not about to give us all a life of infinite leisure any day soon. 

[VWatchie]

  On 2/10/2025 at 12:52 PM, AndyHull said:

Given the state of the world today, I would say the chances of us using it intelligently are probably about 50-50.

Expand  

Like most other technologies.

We had a mass school shooting in Sweden the other day and now our politicians are seriously discussing restrictions for licensed hunters to own hunting weapons. The shooter happened to have licenses for hunting weapons. Like with all other technologies it seems the problem is always with the technology in itself.

I guess I must be stupid as I always thought it was the use of the technology that was the problem. But what do I know. I’m not a politician.

In this context I saw a pretty interesting sign. It read:

”If guns kill people, pencils misspell words, cars drive drunk, and spoons make people fat”.

[RichardHarris123]

Technically kills no-one, we do.  We've been doing it since we appeared as as species and unfortunately will will carry on as long as we exist.  

[JohnR725]

  On 2/10/2025 at 7:40 PM, VWatchie said:

shooting

Expand  

One day I was curious and looked up something and found a really interesting website out there. Can't quite remember how they did this but basically they were looking at how people died. So for instance there appears to be a perception by the rest of the world that the USA is a wild West everybody carries guns and we have shootouts on our streets on a continuous basis. Or basically the number one way of being killed in this country would be with a gun. So the website basically looked at how people died and made an interesting conclusion.

So the simplistic was it's easy to point a finger at somebody died from a gun. Is not so easy to look at non-natural deaths that are achieving the exact same desire of terminating somebody else. West I think going so far as to suggest that non-natural Deaths might actually occur at a higher rate then if you had access to guns. Or basically a lot harder to track.

 

  On 2/10/2025 at 7:40 PM, VWatchie said:

I guess I must be stupid as I always thought it was the use of the technology that was the problem. But what do I know. I’m not a politician.

Expand  

The problem with politicians are they are not very bright and they always take the easy way to do things. So guns are bad if they go away bad things will not occur. Versus some way of spotting when individual or individuals are plotting to do bad things. Often times after bad things occurred and things are being investigated there were indications sometimes that something bad was going to occur and nobody was paying attention. But these solution is to make the bad gun Galway and the problem solved. Or basically politicians will do what they do best which is to make life unpleasant for the rest of us and no We are not going to feel more safe with what they're doing typically.

[AndyHull]

Now that seems like an interesting thing to ask AI.
 

  Quote

Can you estimate the number of firearms produced since 1900 and the average number of people killed per firearm?

Expand  

ChatGPT's reply.
 

  Quote

Estimated Number of Firearms Produced Since 1900

Estimating the total number of firearms produced since 1900 is challenging due to variations in record-keeping and illegal/unregistered weapons, but a reasonable estimate can be made using known production figures:

Current Global Firearm Stock:

The Small Arms Survey (2018) estimated about 1 billion firearms in civilian, military, and law enforcement hands.

Of these, 857 million were civilian-owned.

Historical Firearm Production:

The number of firearms in circulation has increased dramatically since the early 20th century.

The total number of firearms manufactured worldwide since 1900 is estimated to be between 1.5 and 2 billion, considering military conflicts, mass production, and civilian firearm ownership trends.

Average Number of People Killed Per Firearm

To estimate how many people have been killed per firearm:

Estimated Firearm-Related Deaths Since 1900:

Wars (20th & 21st centuries): ~100–150 million deaths (WWI, WWII, civil wars, etc.).

Homicides (global estimates vary, ~250,000 per year recently): ~20–30 million since 1900.

Suicides (~60% of gun deaths in the U.S. and lower elsewhere): ~30–40 million globally.

Other causes (accidental, police, undetermined): ~10–20 million.

Total Estimated Deaths: 150–250 million.

Firearm Deaths Per Firearm:

150–250 million deaths / 1.5–2 billion firearms

Estimated range: 0.1–0.15 people killed per firearm

This suggests that, on average, every 7–10 firearms produced result in one death.

Key Considerations

Many firearms are never used for violence (hunting, sport, collections).

Some firearms are used in multiple deaths (wars, mass shootings, organized crime).

Older firearms are retired, lost, or destroyed, affecting the ratio.

Would you like a breakdown by country or historical period?

Expand  

Now obviously ChatGPT has made some fairly broad assumptions here, and it does clearly make the point that most legally owned civilian guns are not used for violence against other citizens, but it does illustrate the problem of estimating the value versus risk of firearms.

You could argue of course try to make the argument that any manufactured item has  risk of causing death, auto-mobiles being perhaps the obvious one, and that we should not therefore worry about this. The risk is unavoidable.

This kind of misses the point of course. We do have laws about cars and car  use and ownership. We have laws about use of asbestos, sales of tobacco, use off recreational drugs and so forth, and very few people object to those. Gun ownership in the USA is a very strange animal in that sense.

Because the "right to bear arms, and the right to arm bears" is enshrined in the US constitution, it seems to have attained near mythical status. We get the strange argument that the solution to school shootings is "more guns". "We should arm the school security" being one of the more illogical responses.

As you can see, AI, while it can distil the facts to some degree, is no more capable of solving the firearms conundrum than the 8.2 billion humans are.

Perhaps a better more targetted (pun intended) approach would be to ask the more direct question...

  Quote

How many US citizens have been killed by firearms since 1900, how many firearms have been sold in the US since 1900 and what is the ratio of firearms to deaths?

Expand  

This was met with he following response..
 

  Quote

Firearm Deaths in the U.S. Since 1900

Estimating the total number of U.S. citizens killed by firearms since 1900 requires historical data on homicides, suicides, and accidental shootings.

Homicides (1900–2023):

The U.S. has recorded about 750,000–800,000 gun-related homicides since 1900.

Homicide rates peaked in the 1920s (Prohibition era), late 1980s (crack epidemic), and early 1990s.

Suicides (1900–2023):

Firearm suicides have been more common than homicides since the late 20th century.

Total estimated firearm suicides: 1.3–1.5 million.

Accidental Shootings & Other Causes:

Estimated at 100,000–150,000 since 1900.

Total Estimated Firearm Deaths in the U.S. (1900–2023):

2.2 to 2.5 million deaths (including homicides, suicides, and accidents).

Number of Firearms Sold in the U.S. Since 1900

The U.S. has the largest number of civilian-owned firearms in the world.

Total Firearms in Circulation Today:

Estimates suggest over 400 million guns are currently in the U.S.

Historical Firearm Production & Imports (1900–2023):

Estimates suggest 500–600 million firearms have been sold or manufactured in the U.S. since 1900.

This includes mass production during wartime, civilian sales, and imports.

Ratio of Firearms to Firearm Deaths

To find the ratio of firearms sold to deaths:

Total Firearms SoldTotal Firearm Deaths=500 to 600 million2.2 to 2.5 million\frac{\text{Total Firearms Sold}}{\text{Total Firearm Deaths}} = \frac{500 \text{ to } 600 \text{ million}}{2.2 \text{ to } 2.5 \text{ million}}

Estimated ratio: 200–270 firearms per firearm-related death in the U.S. since 1900.

This means that for every 200–270 guns sold or produced, one firearm-related death has occurred.

Key Considerations:

Many guns remain unused for violence (hunting, sport, collections).

Some guns are responsible for multiple deaths, while many never fire a shot outside of practice.

The gun death rate varies significantly over time and by demographic/geographic factors.

Would you like a more detailed breakdown by decade or specific type of firearm death?

Expand  

.. so perhaps this estimate  of one violent death per every two to three hundred guns produced in the USA could be considered a "better" estimate,  but it is still a pretty grim statistic.

A further question ...

  Quote

How many gun shops are there in the US and how many guns does the average gun shop sell in any given year?

Expand  

Yields this response...
 

  Quote

As of September 2017, the United States had approximately 64,417 federally licensed firearm dealers. citeturn0search0 This number includes various types of sellers, such as gun stores, pawnshops, and individual dealers.

In 2022, an estimated 17.4 million firearms were sold in the U.S. citeturn0search1 Dividing this figure by the number of firearm dealers provides an average of approximately 270 firearms sold per dealer in that year.

However, it's important to note that firearm sales are not evenly distributed among dealers. Some high-volume retailers may sell thousands of firearms annually, while smaller shops or individual dealers might sell far fewer. Additionally, these figures can fluctuate based on factors such as changes in legislation, economic conditions, and societal events.

Expand  

.. So statistically speaking, on average, every gun shop in the US is responsible for one violent death per year. Of course this is a highly debatable result, but certainly food for thought.

Edited February 12 by AndyHull