https://www.theverge.com/2024/10/17/24273040/chatgpt-windows-app-subscribers-openai
OpenAI is testing a ChatGPT app for Windows - but it's only available to
paid users for now. You can download an early version of the app from the Microsoft Store.
Just like the Mac version of the app, ChatGPT on Windows lets you ask the AI-powered chatbot questions in a dedicated window that you can keep open alongside your apps. You can quickly access the app by using the Alt +
Space shortcut.
Only ChatGPT Plus, Enterprise, Team, and Edu subscribers can use the app on Windows, OpenAI says it plans on bringing it to everyone later this year.
On 10/17/2024 11:35 PM, Tony T wrote:
https://www.theverge.com/2024/10/17/24273040/chatgpt-windows-app-subscribers-openai
OpenAI is testing a ChatGPT app for Windows - but it's only available to
paid users for now. You can download an early version of the app from the
Microsoft Store.
Just like the Mac version of the app, ChatGPT on Windows lets you ask the
AI-powered chatbot questions in a dedicated window that you can keep open
alongside your apps. You can quickly access the app by using the Alt +
Space shortcut.
Only ChatGPT Plus, Enterprise, Team, and Edu subscribers can use the app on >> Windows, OpenAI says it plans on bringing it to everyone later this year.
Is this functionally any different from using the web version of the product (which you can also keep open in a separate browser window)
https://www.theverge.com/2024/10/17/24273040/chatgpt-windows-app-subscribers-openai
OpenAI is testing a ChatGPT app for Windows - but it's only available to
paid users for now. You can download an early version of the app from the Microsoft Store.
On 10/17/2024 11:35 PM, Tony T wrote:
https://www.theverge.com/2024/10/17/24273040/chatgpt-windows-app-subscribers-openai
OpenAI is testing a ChatGPT app for Windows - but it's only available to
paid users for now. You can download an early version of the app from the
Microsoft Store.
And Google now has an API for their Gemini AI. It's similar
to the maps API, with the ability to spellcheck and "improve"
any writing in any window. Their keylogger is free for personal
use, up to a point. Anyone can use it via script, REST, etc or
via a program designed to access the API.
"AI is a product of the mass surveillance business model in its
current form. It is not a separate technological phenomenon."
- Meredith Whittaker, CEO of Signal and former Google employee https://www.wired.com/story/meredith-whittaker-signal/
On Fri, 10/18/2024 8:18 AM, Newyana2 wrote:
On 10/17/2024 11:35 PM, Tony T wrote:
https://www.theverge.com/2024/10/17/24273040/chatgpt-windows-app-subscribers-openai
OpenAI is testing a ChatGPT app for Windows - but it's only available to >>> paid users for now. You can download an early version of the app from the >>> Microsoft Store.
And Google now has an API for their Gemini AI. It's similar
to the maps API, with the ability to spellcheck and "improve"
any writing in any window. Their keylogger is free for personal
use, up to a point. Anyone can use it via script, REST, etc or
via a program designed to access the API.
"AI is a product of the mass surveillance business model in its
current form. It is not a separate technological phenomenon."
- Meredith Whittaker, CEO of Signal and former Google employee
https://www.wired.com/story/meredith-whittaker-signal/
W-h-a-t c-a-n I m-a-k-e w-i-t-h a p-o-u-n-d o-f H-a-m-b-u-r-g-e-r M-e-a-t ?
https://fortune.com/2024/09/17/oracle-larry-ellison-surveillance-state-police-ai/
"Larry Ellison predicts rise of the modern surveillance state"
https://www.ccn.com/news/technology/ex-google-ceo-eric-schmidt-abandon-climate-targets-let-ai-solve-the-problem/
All of that will be swamped by the enormous needs of this new technology, Schmidt said.
What a time to be alive.
W-h-a-t i-s t-h-e m-o-s-t e-f-f-i-c-i-e-n-t p-a-t-t-y s-h-a-p-e ?
"If you have square bunz, use a square patty.
If you have round bunz, use a round patty.
If you have no bunz, a fractal patty holds the most ketchup and relish."
"There will be a $0.20 charge for this question.
And I want one of your hamburgers too. I like relish
and ketchup on my burger."
Paul wrote:
On Thu, 10/17/2024 11:53 PM, Rick wrote:ChatGPT and CoPilot are two different AI services
On 10/17/2024 11:35 PM, Tony T wrote:
https://www.theverge.com/2024/10/17/24273040/chatgpt-windows-app-subscribers-openaiIs this functionally any different from using the web version of the product (which you can also keep open in a separate browser window)
OpenAI is testing a ChatGPT app for Windows - but it's only available to >>>> paid users for now. You can download an early version of the app from the >>>> Microsoft Store.
Just like the Mac version of the app, ChatGPT on Windows lets you ask the >>>> AI-powered chatbot questions in a dedicated window that you can keep open >>>> alongside your apps. You can quickly access the app by using the Alt + >>>> Space shortcut.
Only ChatGPT Plus, Enterprise, Team, and Edu subscribers can use the app on
Windows, OpenAI says it plans on bringing it to everyone later this year. >>>
This is the icon in the Task Bar (now moved next to start button).
[Picture]
https://i.postimg.cc/cJVZB1N9/copilot-interface-today-win11.gif
When launched fro MSEdge (upper right icon), looks like this instead.
https://i.postimg.cc/mkm5VVHL/launch-copilot-inside-msedge.gif
The prompt box looks the same in each case.
(Win11)
Paul
On 10/18/2024 11:08 AM, Paul wrote:
On Fri, 10/18/2024 8:18 AM, Newyana2 wrote:Why do I get the sense that you like this armageddon of
On 10/17/2024 11:35 PM, Tony T wrote:
https://www.theverge.com/2024/10/17/24273040/chatgpt-windows-app-subscribers-openai
OpenAI is testing a ChatGPT app for Windows - but it's only available to >>>> paid users for now. You can download an early version of the app from the >>>> Microsoft Store.
And Google now has an API for their Gemini AI. It's similar
to the maps API, with the ability to spellcheck and "improve"
any writing in any window. Their keylogger is free for personal
use, up to a point. Anyone can use it via script, REST, etc or
via a program designed to access the API.
"AI is a product of the mass surveillance business model in its
current form. It is not a separate technological phenomenon."
- Meredith Whittaker, CEO of Signal and former Google employee
https://www.wired.com/story/meredith-whittaker-signal/
W-h-a-t c-a-n I m-a-k-e w-i-t-h a p-o-u-n-d o-f H-a-m-b-u-r-g-e-r M-e-a-t ?
https://fortune.com/2024/09/17/oracle-larry-ellison-surveillance-state-police-ai/
"Larry Ellison predicts rise of the modern surveillance state"
https://www.ccn.com/news/technology/ex-google-ceo-eric-schmidt-abandon-climate-targets-let-ai-solve-the-problem/
All of that will be swamped by the enormous needs of this new technology, Schmidt said.
What a time to be alive.
W-h-a-t i-s t-h-e m-o-s-t e-f-f-i-c-i-e-n-t p-a-t-t-y s-h-a-p-e ?
"If you have square bunz, use a square patty.
If you have round bunz, use a round patty.
If you have no bunz, a fractal patty holds the most ketchup and relish." >>
"There will be a $0.20 charge for this question.
And I want one of your hamburgers too. I like relish
and ketchup on my burger."
common sense, intellect and privacy? :)
<https://www.inc.com/jeff-haden/warren-buffett-gets-a-mcmuffin-on-way-to-work-every-morning-but-not-for-reason-you-think.html>
Similarly, Warren Buffet complains in interviews that his
secretary pays more tax than he does. Yet he doesn't do anything
to change the inequity.
Not less taxes than his secretary, but a lower tax rate.
Buffet's Berkshire Hathaway paid $5 Billion in taxes last year(2023). <https://www.youtube.com/watch?v=VJzTsTU1xL8>
Listen to the whole interview(2:46 min-sec) or fast forward to 1:32
The only question is whether Warren Buffet is silly enough to
really believe his farce, desperately trying to remain a regular
guy billionaire, or whether he does it all for show
On 10/19/2024 12:36 PM, ...winston wrote:
Similarly, Warren Buffet complains in interviews that his
secretary pays more tax than he does. Yet he doesn't do anything
to change the inequity.
Not less taxes than his secretary, but a lower tax rate.
Yes, obviously.
Buffet's Berkshire Hathaway paid $5 Billion in taxes last year(2023).
<https://www.youtube.com/watch?v=VJzTsTU1xL8>
Listen to the whole interview(2:46 min-sec) or fast forward to 1:32
$5 billion in taxes means nothing out of context. And I'm afraid
I don't share your fascination with the filthy rich.
"In high school, he invested in a business owned by his father and bought
a 40-acre farm worked by a tenant farmer. He bought the land when he was
14 years old with $1,200 of his savings.
While some things the rich people do are definitely an "act",
some things shine through. The people who are rat bastards,
have always been rat bastards :-) Of that you can be assured.
On 10/20/2024 1:18 AM, Paul wrote:
"In high school, he invested in a business owned by his father and bought
a 40-acre farm worked by a tenant farmer. He bought the land when he was >> 14 years old with $1,200 of his savings.
$1,200 in savings, at 14 y.o.? Which is, what, maybe $12K in
today's dollars? Are we to believe he was working in construction
at 12 and that he grew up in a state with no child labor laws? And
how does a minor buy property? At 14 I had a paper route with
my brother. We each made $5-7 per week. After rewarding ourselves
with a large root beer Slurpee on Saturday mornings when we
cashed in, there was probably enough left to buy a few sodas and
pastries during the week. Or maybe a bicycle tire. I didn't have a job
at 12 (unless you count 25 cents per week for washing the dishes),
so maybe that's why it didn't occur to me to buy a factory
at 14.
While some things the rich people do are definitely an "act",
some things shine through. The people who are rat bastards,
have always been rat bastards :-) Of that you can be assured.
I think the act can take different forms. Much of it is for their
own self-image. It can also be a kind of self-marketing to the
public. It sounds like the "rat bastards" you're talking about are
people who are simply lying to stay out of jail.
In the original article the author was talking a pop psychology
theory that saving on McMuffins is a kind of superstitious way
to attempt controlling the future and/or keep a sharp business
attitude in all things. There could be some of that, but
I suspect that with Warren Buffet it's an attachment to being a
"regular guy", which is also a convenient self-marketing strategy.
He's marketing a folksy persona, to himself as well as to the public.
How do you
become a billionaire and still think that you're a nice guy? You lie
to yourself. Buffet pretends that he's just plain folks. Gates pretends
that he's Mr. Brain, here to spread the benefit of his amazing
intelligence. Jobs was guru to the world. Some people don't really
need to lie to themselves. The fever is all-consuming. But are they
worse people? Gates is more appealing than Trump. But is he less
self-serving or more mature?
We have an American mythology that anyone can get rich if
they're willing to work hard. That myth justifies a king-of-the-hill
culture with no tradition of noblesse oblige. But no one gets rich
through kindness and generosity. No one gets rich through virtue.
One can only get rich on the backs of others. King of the hill is not
the same as equal opportunity. We want to have social classes
without nobless oblige. In that respect, monarchy may be a more
civilized and humane system.
I'd go further and say that anyone that rich has to be highly
neurotic and insecure to be so driven. I have nothing against
Warren Buffet. But I don't buy his hayseed persona. And I
certainly don't think these people should be offered as role
models of a successful life. Hoarding money is not a successful
life. Hoarding money while pretending to give it away is slimy.
I remember a cartoon that I think was by Gahan Wilson. Two
old ladies, dressed in black, sitting on the front porch of a rural
house. The screen door is open and various objects are flying
out the door, up into the sky: A dog, a piano, books...
Caption: "Oh, Harold is taking it with him."
We all understand the humor of that cartoon, yet the fantasy of
being a billionaire carries great allure.
I remember seeing an article somewhere, some years ago, where
Steve Jobs was hanging around with Larry Ellison. They were talking
some kind of business plan and Ellison started calculating how much
profit he could make. Jobs said that he grabbed Ellison by the shoulders, looked him in the eye, and firmly stated, "You don't need any more
money!" It was almost endearing. But of course, Jobs was at least
as driven. He just wanted to be admired for wisdom rather than
money. So he was really just acting superior. Always the guru.
The Tibetans have a saying: "You never know which will come
first, the next day or the next life."
One of the executives queried curiously why a wealthy man like Buffet
picked up the penny.
Buffet's response The beginning of the next billion.
True or not, the point is about the growth of money by making use of
your most valuable asset - time.
On 10/21/2024 12:43 AM, ...winston wrote:
One of the executives queried curiously why a wealthy man like BuffetSo the best use of your time is to work at making money
picked up the penny.
Buffet's response “The beginning of the next billion.”
True or not, the point is about the growth of money by making use of
your most valuable asset - time.
that you don't need? Funny priorities. It's an especially sad
priority for an elderly man at death's door. Wasn't there also
another fairy tale about that? Someone named King Midas,
if I remember correctly.
On 10/21/2024 8:23 AM, Newyana2 wrote:
On 10/21/2024 12:43 AM, ...winston wrote:
One of the executives queried curiously why a wealthy man like BuffetSo the best use of your time is to work at making money
picked up the penny.
Buffet's response “The beginning of the next billion.”
True or not, the point is about the growth of money by making use of
your most valuable asset - time.
that you don't need? Funny priorities. It's an especially sad
priority for an elderly man at death's door. Wasn't there also
another fairy tale about that? Someone named King Midas,
if I remember correctly.
It's his life and his choice. If he enjoys using his skills to make
money, who are you to tell him he should do something different? I
imagine you probably wouldn't want other people to denigrate your choice
of how to use your time - what makes you think Buffet is any different?
So the best use of your time is to work at making money
that you don't need? Funny priorities. It's an especially sad
priority for an elderly man at death's door. Wasn't there also
another fairy tale about that? Someone named King Midas,
if I remember correctly.
It's his life and his choice. If he enjoys using his skills to make
money, who are you to tell him he should do something different? I
imagine you probably wouldn't want other people to denigrate your choice
of how to use your time - what makes you think Buffet is any different?
On 10/21/2024 10:04 AM, Rick wrote:
So the best use of your time is to work at making money
that you don't need? Funny priorities. It's an especially sad
priority for an elderly man at death's door. Wasn't there also
another fairy tale about that? Someone named King Midas,
if I remember correctly.
It's his life and his choice. If he enjoys using his skills to make
money, who are you to tell him he should do something different? I
imagine you probably wouldn't want other people to denigrate your
choice of how to use your time - what makes you think Buffet is any
different?
I'm talking about the whole issue of values. People express
admiration for the man because he's a billionaire. Winston
thought Buffet was teaching valuable lessons. I'm questioning
that.
This is a discussion forum. Rather than shoot the
messenger, maybe you'd like to make your case for why making
far more money than you need would be the best use of your
time? It really gets down to what one thinks is worthwhile in
life. Does life have a purpose? Is Warren Buffet happy? Are you
happy? Is you conscience clear and are your priorities straight if
you die tomorrow? Or perhaps you're one of those people who
believes that "he who dies with the most toys wins" and you
don't want to face any doubts about that view because it makes
you nervous? But who is it that wins? And wins what?
Who am I? I'm someone who endeavors to live consciously, in
accord with my own conscience. That being so, I question
the value of aimless greed. Such questions seem to me like a
good use of valuable time. :)
This subthread actually started with me pointing out that AI
is simply a scam of what Whittaker calls the "survelliance
business model" -- entertaining frippery to get device users
to share every detail of their activities with dataminers. Maybe
you have a thought on that? Perhaps it's your feeling that
since greed is good, then all the spyware is clever and we
should admire these companies for managing to exploit us? But
then, that's fetishizing power rather than greed.
Maybe you think it will be fine when MS imposes their own brand
of AI on all of your computer activities? With any luck, AI may
go out of style, but Copilot could also end up as a kind of Clippy
on steroids, getting into all of your business. What if you can't
shut it off? That's what I was getting at with my initial post --
pointing out that despite all the talk about AI, it's essentially
just a frontend for data collection.
Maybe you think it will be fine when MS imposes their own brand
of AI on all of your computer activities? With any luck, AI may
go out of style, but Copilot could also end up as a kind of Clippy
on steroids, getting into all of your business. What if you can't
shut it off? That's what I was getting at with my initial post --
pointing out that despite all the talk about AI, it's essentially
just a frontend for data collection.
The point is that it is up to each person to decide what they want to do
with their life and what makes them happy. As long as they don't commit force against me or take my possessions from me
Are you looking at the GUI for that ?
It's pretty funny actually.
This is the CoPilot icon to the right of the Start icon on Win11 22631.5417 23H2.
The old interface is still available, in MSEdge browser, upper right corner.
You can paste a text question in there, such as my "write me a Hello World program in C". I wanted to do a test to see if that "skill" was in the machine.
I had previously asked it to draw a picture, which the machine claimed was
a subset of CoPilot Studio, and while the drawing concept was good,
the execution left a lot to be desired. And there is an ecosystem of
"cleanup AI tools" which attempt to take the output from a competitor,
and remove the "noise" from the picture.
But one thing you *cannot* do with the AI, is you cannot say
"Tell me your capabilities", for that is an unbounded question,
the machine will suffer a "timeout" (as the cooling water is starting
to boil now), and the machine will erase the screen as a way of proving
it's a timeout. So instead, you have to pose a test question first, keep
it simple (so the machine can just copy it from somewhere), and that's
your proof of a capability.
To see if you're any good at manipulating an AI, see if you
can make it hum the lyrics of a popular tune... in *your* voice :-)
They were supposed to add a quality control to the output, to stop
that from happening, but I think someone got past that.
On 10/21/2024 2:35 PM, Rick wrote:
The point is that it is up to each person to decide what they want to
do with their life and what makes them happy. As long as they don't
commit force against me or take my possessions from me
You've nicely stated the position of modern moral relativism.
We're all here to "get mine" so long as we don't impede someone
else getting theirs. And naturally you feel insulted and imposed
upon by the very idea of ethics. Ethics are immoral! Breathtaking.
I can't say that I have any curiosity. It's sort of interesting
that it can do complex tasks, but the target of complex *useful*
tasks seems elusive as long as we're talking about AI in the role
of Man Friday. I was reading today that OpenAI is having
troubles and chasing 10s of billions in investment, while they
lose billions. So maybe this fad is running its course.
It costs something like a billion dollars per year, to scrape the
Internet for search engine purposes. And that likely did not
require "paying" for content. Whereas the scraping done for
training an AI, is now starting to cost money (for example,
scraping Reddit would involve a separate contract, scraping
arstechnica is another contract and so on).
I really wanted to keep the
voluminous output, the stuff that got erased.
How many letter "r" are present in the word Strawberry ?
The answer to that is now *hardwired* into the AI :-)
Ask the question five times in a row. Sometimes
the count is 2, sometimes the count is 3. Down
at the berry farm, we've used up all our tokens
testing this.
What is breathtaking is how presumptive you are in assuming you
understand my view towards ethics. As a society we can impose the legal restriction that people not be permitted to commit force or fraud
against others, but within that constraint, ethics are in the eye of the beholder and up to each person to decide. I certainly have my own code
of ethics and am neither insulted by the concept nor do I regard them as immoral.
Buffet aside, I could have posted about Mr. R. Read - born into a
indigent farming family, WWII veteran, retired gas station attendant,
J.C. Penney janitor, drove a second hand car, cut his own firewood, and
used safety pins to hold his worn out coat together. Blue-chip value. blue-chip smarts and lifestyle. Left an estate valued over 8 million
much from investments in Wells Fargo, Proctor and Gamble,
Colgate-Palmolive and other companies(incidentally some the same
holdings as Buffet). Bequeathed $2M to his stepchildren and care-givers,
$4M to a hospital, $1.2M to a library. Often freqented a local cafeteria
for the same breakfast - coffee and an English muffin with peanut butter.
As i see it...it's not about values, lessons, ethics, moralism...what
seems clearer than anything else is a complete misconception that fine, outstanding folks having salt of the earth origins know how build wealth using their best asset(time) and in the end some even paying it forward.
But since this is a Windows newsgroup and you like to complain about
Windows and Microsoft so often....here's another statistic for you start another complaint.
10 shares of MSFT stock at ip in 1986(for a whopping total of $210) at current price is worth $1,206,086. (1 single share at $21 - $12,061)
Who knows, even you may have had the foresight at that time to buy 1 share.
On 10/21/2024 10:01 PM, Rick wrote:
What is breathtaking is how presumptive you are in assuming you
understand my view towards ethics. As a society we can impose the
legal restriction that people not be permitted to commit force or
fraud against others, but within that constraint, ethics are in the
eye of the beholder and up to each person to decide. I certainly have
my own code of ethics and am neither insulted by the concept nor do I
regard them as immoral.
I understand because you're telling me. Perhaps you don't
understand your own view on ethics? You just said it's "in the
eye of the beholder". You keep defining a moral relativism with
no set definition of virtue and vice. You're defining regulated
barbarism: Grab what you can so long as you don't grab it from
someone else, especially if that someone else has lawyers.
Ethics is not law. It's about having moral guidelines that
make sense to you, that you try to live by. It's about trying
to live in such a way that you're at peace with your own
conscience, trying to cultivate virtue and lessen vice. To accept
greed as a proper focus of one's energies is simply unethical.
Why? Because greed is selfish. It's indulgence in primitive
instincts. It harms others. There's a reason that virtue is virtue
and vice is vice. It's not in the eye of the beholder. Virtue is
that which is conducive to noble human qualities such as
honestly, generosity, kindness, etc. Vice is that which intensifies
selfish motives.
I see reactions like yours a lot and find it very odd. It seems
to be especially common with younger people. Someone rants
about what they believe, and criticizes others, then when they're
questioned about their statements they blurt out, "Hey, screw you!
You don't know what I think!" Well... yes... actually I do. Because
you specifically posted what you think.
This seems to be connected with the presumption of anonymity
online. People feel they can say anything and shouldn't have to
answer for it. It seems to be especially prevalent with online cancel
and blaming culture. People gossip and attack others mercilessly,
but then when held to account they protest, "Hey you don't know
me. Shut up!"
Really? It's not even accurate with basic technical data?
I suppose maybe that shouldn't be surprising, since it's designed
to imitate intelligence by synthesizing new data. But it's still
just algorythms. If, then, else. So the "truth" it synthesizes
will necessarily be based on random textthat's been run through
various functions, which the programmers have lost track of.
I suppose it wouldmn't be surprising to find that every 5th
response is "lorem ipsum", since sample text often uses that
as filler.
On Tue, 10/22/2024 7:47 AM, Newyana2 wrote:
Really? It's not even accurate with basic technical data?
I suppose maybe that shouldn't be surprising, since it's designed
to imitate intelligence by synthesizing new data. But it's still
just algorythms. If, then, else. So the "truth" it synthesizes
will necessarily be based on random textthat's been run through
various functions, which the programmers have lost track of.
I suppose it wouldmn't be surprising to find that every 5th
response is "lorem ipsum", since sample text often uses that
as filler.
You might notice in the news, that a lot of people use these AI
for writing summaries of presentations in auditoriums. You can
point your smartphone at the projector screen, lay back and take
a snooze, and when you wake up, the fifty slides the presenter
put up, have been summarized into three sentences. People hired
as reporters, love this feature.
As a result, lorem ipsum remains a "concept". The AI is fully
capable of saying to you "it's like lorem ipsum", without
actually dumping the contents of lorem ipsum. It would only
give the verbatim text if you ask it to "give me the exact
text of lorem ipsum". Then it is constrained to no longer
discuss at the concept level.
And you have to think about what you're asking for. PI is an
irrational number, and the output goes on forever. There are
easily online references with four billion digits. You don't
want the AI attempting to give you all the digits it's got.
In the old programmable calculator days, getting fifty digits
of PI was "a miracle". With the AI today, it could easily give
you four billion worth, not by calculating them, but by referencing
some collection of digits it's been shown.
"Give me the first fifty digits of PI"
as your question, prevents boiling the cooling water at the data center.
It was the same with data bases in the past. On our database
at work, when you logged in, there was a message of the day
at the bottom that stated "Do NOT light up the entire database.
Doing so, you will lose your privileges to use it." And instead
of asking for 157,000 transistor datasheets, you had to be
a bit more... careful. It's the same with the AI. If you asked
for the entire database, the disk drive for it was quite slow,
and it would take a beating coughing up an answer, and it
affected how quickly the queries being done by other people,
got executed. One way to unfairly use a computer, is
to busy out the disk drive (on purpose of course). You could get
your account yanked for that on the mainframe.
Paul
On 10/22/2024 11:06 AM, Paul wrote:or third time, but every time I have tested this, the list is not accurate.
On Tue, 10/22/2024 7:47 AM, Newyana2 wrote:
Really? It's not even accurate with basic technical data?
I suppose maybe that shouldn't be surprising, since it's designed
to imitate intelligence by synthesizing new data. But it's still
just algorythms. If, then, else. So the "truth" it synthesizes
will necessarily be based on random textthat's been run through
various functions, which the programmers have lost track of.
I suppose it wouldmn't be surprising to find that every 5th
response is "lorem ipsum", since sample text often uses that
as filler.
You might notice in the news, that a lot of people use these AI
for writing summaries of presentations in auditoriums. You can
point your smartphone at the projector screen, lay back and take
a snooze, and when you wake up, the fifty slides the presenter
put up, have been summarized into three sentences. People hired
as reporters, love this feature.
As a result, lorem ipsum remains a "concept". The AI is fully
capable of saying to you "it's like lorem ipsum", without
actually dumping the contents of lorem ipsum. It would only
give the verbatim text if you ask it to "give me the exact
text of lorem ipsum". Then it is constrained to no longer
discuss at the concept level.
And you have to think about what you're asking for. PI is an
irrational number, and the output goes on forever. There are
easily online references with four billion digits. You don't
want the AI attempting to give you all the digits it's got.
In the old programmable calculator days, getting fifty digits
of PI was "a miracle". With the AI today, it could easily give
you four billion worth, not by calculating them, but by referencing
some collection of digits it's been shown.
"Give me the first fifty digits of PI"
as your question, prevents boiling the cooling water at the data center.
It was the same with data bases in the past. On our database
at work, when you logged in, there was a message of the day
at the bottom that stated "Do NOT light up the entire database.
Doing so, you will lose your privileges to use it." And instead
of asking for 157,000 transistor datasheets, you had to be
a bit more... careful. It's the same with the AI. If you asked
for the entire database, the disk drive for it was quite slow,
and it would take a beating coughing up an answer, and it
affected how quickly the queries being done by other people,
got executed. One way to unfairly use a computer, is
to busy out the disk drive (on purpose of course). You could get
your account yanked for that on the mainframe.
Paul
And yet ChatGPT, at least in its free form, has odd weaknesses. Go to the ChatGPT website and ask this question:
"List the US presidents in reverse chronological order of their age at death or their current age if still living."
I have asked ChatGPT this several times and have never once gotten an accurate list. Yes, the first several entries may look correct, but the complete list will have errors. I have pointed this out to the AI and it has apologized and tried a second
OTOH, I can ask ChatGPT which has the greater effect on the accuracy of GPS satellites, the Special Relativity effect of time slowing down for objects in relative motion compared to objects at rest on earth or the General Relativity effect that timespeeds up for objects away from the earth compared to objects on the planet, and it not only understood what i was asking, but it gave me a mathematically correct analysis.
Go figure.
On Tue, 10/22/2024 12:33 PM, Rick wrote:second or third time, but every time I have tested this, the list is not accurate.
On 10/22/2024 11:06 AM, Paul wrote:
On Tue, 10/22/2024 7:47 AM, Newyana2 wrote:
Really? It's not even accurate with basic technical data?
I suppose maybe that shouldn't be surprising, since it's designed
to imitate intelligence by synthesizing new data. But it's still
just algorythms. If, then, else. So the "truth" it synthesizes
will necessarily be based on random textthat's been run through
various functions, which the programmers have lost track of.
I suppose it wouldmn't be surprising to find that every 5th
response is "lorem ipsum", since sample text often uses that
as filler.
You might notice in the news, that a lot of people use these AI
for writing summaries of presentations in auditoriums. You can
point your smartphone at the projector screen, lay back and take
a snooze, and when you wake up, the fifty slides the presenter
put up, have been summarized into three sentences. People hired
as reporters, love this feature.
As a result, lorem ipsum remains a "concept". The AI is fully
capable of saying to you "it's like lorem ipsum", without
actually dumping the contents of lorem ipsum. It would only
give the verbatim text if you ask it to "give me the exact
text of lorem ipsum". Then it is constrained to no longer
discuss at the concept level.
And you have to think about what you're asking for. PI is an
irrational number, and the output goes on forever. There are
easily online references with four billion digits. You don't
want the AI attempting to give you all the digits it's got.
In the old programmable calculator days, getting fifty digits
of PI was "a miracle". With the AI today, it could easily give
you four billion worth, not by calculating them, but by referencing
some collection of digits it's been shown.
"Give me the first fifty digits of PI"
as your question, prevents boiling the cooling water at the data center. >>>
It was the same with data bases in the past. On our database
at work, when you logged in, there was a message of the day
at the bottom that stated "Do NOT light up the entire database.
Doing so, you will lose your privileges to use it." And instead
of asking for 157,000 transistor datasheets, you had to be
a bit more... careful. It's the same with the AI. If you asked
for the entire database, the disk drive for it was quite slow,
and it would take a beating coughing up an answer, and it
affected how quickly the queries being done by other people,
got executed. One way to unfairly use a computer, is
to busy out the disk drive (on purpose of course). You could get
your account yanked for that on the mainframe.
Paul
And yet ChatGPT, at least in its free form, has odd weaknesses. Go to the ChatGPT website and ask this question:
"List the US presidents in reverse chronological order of their age at death or their current age if still living."
I have asked ChatGPT this several times and have never once gotten an accurate list. Yes, the first several entries may look correct, but the complete list will have errors. I have pointed this out to the AI and it has apologized and tried a
speeds up for objects away from the earth compared to objects on the planet, and it not only understood what i was asking, but it gave me a mathematically correct analysis.
OTOH, I can ask ChatGPT which has the greater effect on the accuracy of GPS satellites, the Special Relativity effect of time slowing down for objects in relative motion compared to objects at rest on earth or the General Relativity effect that time
Go figure.
I would start with:
1) List all the US Presidents.
2) List all the US Presidents and their birth date.
Don't specify a sort order. The AI can do that
themselves when writing the summary.
Some historical documents, would be OCR-converted from scans,
and not all the OCR ever done, was of good quality.
When you asked for your list, there were web links at the
bottom of the answer. Were any of those damaged ?
In the olden days, birth registry was a slow, error filled process.
And a celebrity figure, leave it to historians to muddy the waters.
For example, my fathers name was mis-spelled on the birth certificate.
After he passed away, my sister had that fixed! That's an attempt
to make all the paperwork align.
Paul
Paul wrote:
On Tue, 10/22/2024 12:33 PM, Rick wrote:
On 10/22/2024 11:06 AM, Paul wrote:
On Tue, 10/22/2024 7:47 AM, Newyana2 wrote:
Really? It's not even accurate with basic technical data?
I suppose maybe that shouldn't be surprising, since it's designed
to imitate intelligence by synthesizing new data. But it's still
just algorythms. If, then, else. So the "truth" it synthesizes
will necessarily be based on random textthat's been run through
various functions, which the programmers have lost track of.
I suppose it wouldmn't be surprising to find that every 5th
response is "lorem ipsum", since sample text often uses that
as filler.
You might notice in the news, that a lot of people use these AI
for writing summaries of presentations in auditoriums. You can
point your smartphone at the projector screen, lay back and take
a snooze, and when you wake up, the fifty slides the presenter
put up, have been summarized into three sentences. People hired
as reporters, love this feature.
As a result, lorem ipsum remains a "concept". The AI is fully
capable of saying to you "it's like lorem ipsum", without
actually dumping the contents of lorem ipsum. It would only
give the verbatim text if you ask it to "give me the exact
text of lorem ipsum". Then it is constrained to no longer
discuss at the concept level.
And you have to think about what you're asking for. PI is an
irrational number, and the output goes on forever. There are
easily online references with four billion digits. You don't
want the AI attempting to give you all the digits it's got.
In the old programmable calculator days, getting fifty digits
of PI was "a miracle". With the AI today, it could easily give
you four billion worth, not by calculating them, but by referencing
some collection of digits it's been shown.
"Give me the first fifty digits of PI"
as your question, prevents boiling the cooling water at the data
center.
It was the same with data bases in the past. On our database
at work, when you logged in, there was a message of the day
at the bottom that stated "Do NOT light up the entire database.
Doing so, you will lose your privileges to use it." And instead
of asking for 157,000 transistor datasheets, you had to be
a bit more... careful. It's the same with the AI. If you asked
for the entire database, the disk drive for it was quite slow,
and it would take a beating coughing up an answer, and it
affected how quickly the queries being done by other people,
got executed. One way to unfairly use a computer, is
to busy out the disk drive (on purpose of course). You could get
your account yanked for that on the mainframe.
Paul
And yet ChatGPT, at least in its free form, has odd weaknesses. Go
to the ChatGPT website and ask this question:
"List the US presidents in reverse chronological order of their age
at death or their current age if still living."
I have asked ChatGPT this several times and have never once gotten an
accurate list. Yes, the first several entries may look correct, but
the complete list will have errors. I have pointed this out to the
AI and it has apologized and tried a second or third time, but every
time I have tested this, the list is not accurate.
OTOH, I can ask ChatGPT which has the greater effect on the accuracy
of GPS satellites, the Special Relativity effect of time slowing down
for objects in relative motion compared to objects at rest on earth
or the General Relativity effect that time speeds up for objects away
from the earth compared to objects on the planet, and it not only
understood what i was asking, but it gave me a mathematically correct
analysis.
Go figure.
I would start with:
1) List all the US Presidents.
2) List all the US Presidents and their birth date.
Don't specify a sort order. The AI can do that
themselves when writing the summary.
Some historical documents, would be OCR-converted from scans,
and not all the OCR ever done, was of good quality.
When you asked for your list, there were web links at the
bottom of the answer. Were any of those damaged ?
In the olden days, birth registry was a slow, error filled process.
And a celebrity figure, leave it to historians to muddy the waters.
For example, my fathers name was mis-spelled on the birth certificate.
After he passed away, my sister had that fixed! That's an attempt
to make all the paperwork align.
Paul
Exclude 'list all the'
CoPilot
us presidents birth and death dates
Here's a list of U.S. Presidents along with their birth and death dates:
President Birth Date Death Date
George Washington Feb 22, 1732 Dec 14, 1799
John Adams Oct 30, 1735 July 4, 1826
Thomas Jefferson Apr 13, 1743 July 4, 1826
James Madison Mar 16, 1751 June 28, 1836
James Monroe Apr 28, 1758 July 4, 1831
John Quincy Adams July 11, 1767 Feb 23, 1848
Andrew Jackson Mar 15, 1767 June 8, 1845
Martin Van Buren Dec 5, 1782 July 24, 1862
William Henry Harrison Feb 9, 1773 Apr 4, 1841
John Tyler Mar 29, 1790 Jan 18, 1862
James K. Polk Nov 2, 1795 June 15, 1849
Zachary Taylor Nov 24, 1784 July 9, 1850
Millard Fillmore Jan 7, 1800 Mar 8, 1874
Franklin Pierce Nov 23, 1804 Oct 8, 1869
James Buchanan Apr 23, 1791 June 1, 1868
Abraham Lincoln Feb 12, 1809 Apr 15, 1865
Andrew Johnson Dec 29, 1808 July 31, 1875
Ulysses S. Grant Apr 27, 1822 July 23, 1885
Rutherford B. Hayes Oct 4, 1822 Jan 17, 1893
James A. Garfield Nov 19, 1831 Sep 19, 1881
Chester Arthur Oct 5, 1829 Nov 18, 1886
Grover Cleveland Mar 18, 1837 June 24, 1908
Benjamin Harrison Aug 20, 1833 Mar 13, 1901
William McKinley Jan 29, 1843 Sep 14, 1901
Theodore Roosevelt Oct 27, 1858 Jan 6, 1919
William Howard Taft Sep 15, 1857 Mar 8, 1930
Woodrow Wilson Dec 28, 1856 Feb 3, 1924
Warren G. Harding Nov 2, 1865 Aug 2, 1923
Calvin Coolidge July 4, 1872 Jan 5, 1933
Herbert Hoover Aug 10, 1874 Oct 20, 1964
Franklin D. Roosevelt Jan 30, 1882 Apr 12, 1945
Harry S. Truman May 8, 1884 Dec 26, 1972
Dwight Eisenhower Oct 14, 1890 Mar 28, 1969
John F. Kennedy May 29, 1917 Nov 22, 1963
Lyndon B. Johnson Aug 27, 1908 Jan 22, 1973
Richard Nixon Jan 9, 1913 Apr 22, 1994
Gerald Ford July 14, 1913 Dec 26, 2006
Jimmy Carter Oct 1, 1924 (still living)
Ronald Reagan Feb 6, 1911 June 5, 2004
George H. W. Bush Jun 12, 1924 Nov 30, 2018
Bill Clinton Aug 19, 1946 (still living)
George W. Bush Jul 6, 1946 (still living)
Barack Obama Aug 4, 1961 (still living)
Donald Trump Jun 14, 1946 (still living)
Joe Biden Nov 20, 1942 (still living)
Is there a specific president you're interested in learning more about?
On 10/22/2024 8:46 PM, Rick wrote:
But that wasn't the original task. The original task was to list the
presidents (living and dead) in reverse chronological order of their
age at death or their current age if still alive. In the case of
ChatGPT, it was able to get the correct information on each
president's age at death or current age if alive, but it failed to
properly sequence them, even after several attempts. I just found
this to be an odd deficiency in light of all the things it can do
correctly.
I don't think correct is a relevant parameter. It tries to guess what
you want and then produces that with whatever data it has. And
much of that data is scraped from websites. Truth is not relevant at
any point in the process.
There was a news story awhile back that demonstrated that nicely.
A lawyer got in trouble in a court case. He had used AI to research
precedent relating to the case and the AI came up with 6 cases, which
he then presented in court. But all 6 were complete fabrications. The
AI gave the lawyer what he asked for.
But that wasn't the original task. The original task was to list the presidents (living and dead) in reverse chronological order of their age
at death or their current age if still alive. In the case of ChatGPT,
it was able to get the correct information on each president's age at
death or current age if alive, but it failed to properly sequence them,
even after several attempts. I just found this to be an odd deficiency
in light of all the things it can do correctly.
List all the US Presidents
On 10/22/2024 11:24 PM, Newyana2 wrote:information on each president's age at death or current age if alive, but it failed to properly sequence them, even after several attempts. I just found this to be an odd deficiency in light of all the things it can do correctly.
On 10/22/2024 8:46 PM, Rick wrote:
But that wasn't the original task. The original task was to list the presidents (living and dead) in reverse chronological order of their age at death or their current age if still alive. In the case of ChatGPT, it was able to get the correct
I don't think correct is a relevant parameter. It tries to guess what >> you want and then produces that with whatever data it has. And
much of that data is scraped from websites. Truth is not relevant at
any point in the process.
There was a news story awhile back that demonstrated that nicely.
A lawyer got in trouble in a court case. He had used AI to research
precedent relating to the case and the AI came up with 6 cases, which
he then presented in court. But all 6 were complete fabrications. The
AI gave the lawyer what he asked for.
But the point is that it did have the correct data. It correctly listed all the presidents with their correct ages at death. It did everything right except it couldn't do the simple task of arranging them in the requested order.
And it wasn't even a case of not understanding the request. It clearly produced a list that was mostly in the requested sequence. But it just wasn't quite complete.
And it wasn't even a case of not understanding the request. It clearly produced a list that was mostly in the requested sequence. But it just wasn't quite complete.
It's my suspicion, you ran out of tokens.
You may have hit a problem size limitation,
but I don't know of a way to research that aspect.
On Tue, 10/22/2024 11:43 PM, Rick wrote:information on each president's age at death or current age if alive, but it failed to properly sequence them, even after several attempts. I just found this to be an odd deficiency in light of all the things it can do correctly.
On 10/22/2024 11:24 PM, Newyana2 wrote:
On 10/22/2024 8:46 PM, Rick wrote:
But that wasn't the original task. The original task was to list the presidents (living and dead) in reverse chronological order of their age at death or their current age if still alive. In the case of ChatGPT, it was able to get the correct
I don't think correct is a relevant parameter. It tries to guess what >>> you want and then produces that with whatever data it has. And
much of that data is scraped from websites. Truth is not relevant at
any point in the process.
There was a news story awhile back that demonstrated that nicely.
A lawyer got in trouble in a court case. He had used AI to research
precedent relating to the case and the AI came up with 6 cases, which
he then presented in court. But all 6 were complete fabrications. The
AI gave the lawyer what he asked for.
But the point is that it did have the correct data. It correctly listed all the presidents with their correct ages at death. It did everything right except it couldn't do the simple task of arranging them in the requested order.
And it wasn't even a case of not understanding the request. It clearly produced a list that was mostly in the requested sequence. But it just wasn't quite complete.
It's my suspicion, you ran out of tokens.
You may have hit a problem size limitation,
but I don't know of a way to research that aspect.
Paul
On 10/23/2024 3:31 PM, Paul wrote:information on each president's age at death or current age if alive, but it failed to properly sequence them, even after several attempts. I just found this to be an odd deficiency in light of all the things it can do correctly.
On Tue, 10/22/2024 11:43 PM, Rick wrote:
On 10/22/2024 11:24 PM, Newyana2 wrote:
On 10/22/2024 8:46 PM, Rick wrote:
But that wasn't the original task. The original task was to list the presidents (living and dead) in reverse chronological order of their age at death or their current age if still alive. In the case of ChatGPT, it was able to get the correct
omitted Barack Obama who was clearly born after 1900. And the 11 names it listed, while partially in order for the first four entries, had errors farther down the list.
I don't think correct is a relevant parameter. It tries to guess what
you want and then produces that with whatever data it has. And
much of that data is scraped from websites. Truth is not relevant at
any point in the process.
There was a news story awhile back that demonstrated that nicely. >>>> A lawyer got in trouble in a court case. He had used AI to research
precedent relating to the case and the AI came up with 6 cases, which
he then presented in court. But all 6 were complete fabrications. The
AI gave the lawyer what he asked for.
But the point is that it did have the correct data. It correctly listed all the presidents with their correct ages at death. It did everything right except it couldn't do the simple task of arranging them in the requested order.
And it wasn't even a case of not understanding the request. It clearly produced a list that was mostly in the requested sequence. But it just wasn't quite complete.
It's my suspicion, you ran out of tokens.
You may have hit a problem size limitation,
but I don't know of a way to research that aspect.
Paul
Well to narrow the scope, I asked it to list only the presidents born after 1900 in reverse chronological order of age at death or current age if alive. That reduced the list to only 12 presidents. Yet it couldn't even get this right. First it
Here is the exact output I received:
U.S. Presidents Born After 1900, Ordered by Age at Death or Current Age:
1. Jimmy Carter
Born: October 1, 1924
Current Age: 100 years, 22 days (As of October 23, 2024)
2. George H. W. Bush
Born: June 12, 1924
Died: November 30, 2018 (Age: 94 years, 171 days)
3. Gerald Ford
Born: July 14, 1913
Died: December 26, 2006 (Age: 93 years, 165 days)
4. Ronald Reagan
Born: February 6, 1911
Died: June 5, 2004 (Age: 93 years, 120 days)
5. Richard Nixon
Born: January 9, 1913
Died: April 22, 1994 (Age: 81 years, 103 days)
6. Lyndon B. Johnson
Born: August 27, 1908
Died: January 22, 1973 (Age: 64 years, 148 days)
7. John F. Kennedy
Born: May 29, 1917
Died: November 22, 1963 (Assassinated at Age: 46 years, 177 days)
8. George W. Bush
Born: July 6, 1946
Current Age: 78 years, 109 days
9. Bill Clinton
Born: August 19, 1946
Current Age: 78 years, 65 days
10. Donald Trump
Born: June 14, 1946
Current Age: 78 years, 131 days
11. Joe Biden
Born: November 20, 1942
Current Age: 81 years, 338 days
This list includes all U.S. Presidents born after 1900, arranged by their age at death (if deceased) or their current age if still alive.
On 10/23/2024 3:31 PM, Paul wrote:
And it wasn't even a case of not understanding the request. It clearly produced a list that was mostly in the requested sequence. But it just wasn't quite complete.
It's my suspicion, you ran out of tokens.
You may have hit a problem size limitation,
but I don't know of a way to research that aspect.
That's an interesting aspect of this. As you detailed earlier,
ChatGPT is failing at a task that could easily be done with a
sorting routine. But there isn't really any way to backtrack and
figure out what sorting routine was used -- or what other
factors played into it. They're trying to engineer intelligence
but all they have is incredibly complex programming code that
they've lost track of.
On Wed, 10/23/2024 5:55 PM, Rick wrote:information on each president's age at death or current age if alive, but it failed to properly sequence them, even after several attempts. I just found this to be an odd deficiency in light of all the things it can do correctly.
On 10/23/2024 3:31 PM, Paul wrote:
On Tue, 10/22/2024 11:43 PM, Rick wrote:
On 10/22/2024 11:24 PM, Newyana2 wrote:
On 10/22/2024 8:46 PM, Rick wrote:
But that wasn't the original task. The original task was to list the presidents (living and dead) in reverse chronological order of their age at death or their current age if still alive. In the case of ChatGPT, it was able to get the correct
omitted Barack Obama who was clearly born after 1900. And the 11 names it listed, while partially in order for the first four entries, had errors farther down the list.
I don't think correct is a relevant parameter. It tries to guess what
you want and then produces that with whatever data it has. And
much of that data is scraped from websites. Truth is not relevant at >>>>> any point in the process.
There was a news story awhile back that demonstrated that nicely. >>>>> A lawyer got in trouble in a court case. He had used AI to research
precedent relating to the case and the AI came up with 6 cases, which >>>>> he then presented in court. But all 6 were complete fabrications. The >>>>> AI gave the lawyer what he asked for.
But the point is that it did have the correct data. It correctly listed all the presidents with their correct ages at death. It did everything right except it couldn't do the simple task of arranging them in the requested order.
And it wasn't even a case of not understanding the request. It clearly produced a list that was mostly in the requested sequence. But it just wasn't quite complete.
It's my suspicion, you ran out of tokens.
You may have hit a problem size limitation,
but I don't know of a way to research that aspect.
Paul
Well to narrow the scope, I asked it to list only the presidents born after 1900 in reverse chronological order of age at death or current age if alive. That reduced the list to only 12 presidents. Yet it couldn't even get this right. First it
Here is the exact output I received:
U.S. Presidents Born After 1900, Ordered by Age at Death or Current Age:
1. Jimmy Carter
Born: October 1, 1924
Current Age: 100 years, 22 days (As of October 23, 2024)
2. George H. W. Bush
Born: June 12, 1924
Died: November 30, 2018 (Age: 94 years, 171 days)
3. Gerald Ford
Born: July 14, 1913
Died: December 26, 2006 (Age: 93 years, 165 days)
4. Ronald Reagan
Born: February 6, 1911
Died: June 5, 2004 (Age: 93 years, 120 days)
5. Richard Nixon
Born: January 9, 1913
Died: April 22, 1994 (Age: 81 years, 103 days)
6. Lyndon B. Johnson
Born: August 27, 1908
Died: January 22, 1973 (Age: 64 years, 148 days)
7. John F. Kennedy
Born: May 29, 1917
Died: November 22, 1963 (Assassinated at Age: 46 years, 177 days)
8. George W. Bush
Born: July 6, 1946
Current Age: 78 years, 109 days
9. Bill Clinton
Born: August 19, 1946
Current Age: 78 years, 65 days
10. Donald Trump
Born: June 14, 1946
Current Age: 78 years, 131 days
11. Joe Biden
Born: November 20, 1942
Current Age: 81 years, 338 days
This list includes all U.S. Presidents born after 1900, arranged by their age at death (if deceased) or their current age if still alive.
What if you take the method, and state it in multiple sentences.
Similar to this, but you adjust for what you want:
Create a list for yourself, of the US Presidents, their Birth Date, their age at death (if dead), their current age (if alive)
(This means, for each US President in the list, some "age" value will be present, either age at death or current age.)
Remove from the list, US Presidents who were born before the year 1900 .
Sort the remaining list items using the "age" field.
Then, present the list, checking carefully that the sort operation did not lose any entries.
This is just a test run, attempting to make the sort operation as easy as is possible.
I didn't want to throw in reverse chronological, on the first test run.
Breaking it down into steps, may not transfer the result to "temporary storage",
so in the end, there's no particular reason for this to work. Like, inserting a line "work step by step" or "check your work", there may already be prompts like that in the program they use.
You would think, with the layers of quality control that are
supposed to be present, the stupid thing couldn't lose an entry.
Paul
| Sysop: | Keyop |
|---|---|
| Location: | Huddersfield, West Yorkshire, UK |
| Users: | 361 |
| Nodes: | 16 (2 / 14) |
| Uptime: | 123:24:14 |
| Calls: | 7,716 |
| Files: | 12,861 |
| Messages: | 5,727,955 |
