ChatGPT is a gibbering idiot (this is not news)

А как мозг узнаёт, какие именно синапсы задействовать при обдумывании той или иной задачи? С ИИ понятно - он видит знакомое слово и по нему ищет уже имеющуюся в памяти информацию. Он не анализирует, он сравнивает и сопоставляет. А человек сам генерирует результат, анализирует и делает выводы.
I suspect that at quantum level information processing becomes much simpler than at macro size.
IMO, at that level complexity is often elementary fractal, like fields with emergent patterns. But that is outside my level of stored information in memory...1746875092138.png

I know its drifting off-topic.. I'll stop now.
 
I suspect that at quantum level information processing becomes much simpler than at macro size.
IMO, at that level complexity is often elementary fractal, like fields with emergent patterns. But that is outside my level of stored information in memory...View attachment 6771

I know its drifting off-topic.. I'll stop now.
Я думаю, законы физики(вернее, закон) одинаковы и на макро, и на микроуровнях.
 
I think the laws of physics (or rather, the law) are the same at both the macro and micro levels.
Funny you say that, because so far, it seems they are not.

The of the biggest bones stuck in the craw of physics today is that quantum mechanics - the atomic realm - and general relatibty - the cosmological realm - are currently incompatible.

One operates as a field, meaning every point, even infinitesimally close, has a value, whereas the other tells us that can't be so or we get infinities popping up all over the place.

The two greatest, and most thoroughly tested, theories in the history of science are incompatible.

We'll sort it out, but currently it's a big problem.
 
Funny you say that, because so far, it seems they are not.

The of the biggest bones stuck in the craw of physics today is that quantum mechanics - the atomic realm - and general relatibty - the cosmological realm - are currently incompatible.

One operates as a field, meaning every point, even infinitesimally close, has a value, whereas the other tells us that can't be so or we get infinities popping up all over the place.

The two greatest, and most thoroughly tested, theories in the history of science are incompatible.

We'll sort it out, but currently it's a big problem.
Нет никакой в этом никакой проблемы, Дэйв. И никаких полей, тоже нет(ну, если только ромашковые, или пшеничные, или кукурузные, или футбольные, или ещё что то типа того). Просто каждая частица создаёт свой тип колебания пространства, и вы принимаете это за возмущение полей. Я вообще то уже это объясняла немного, но вы, похоже, даже и не читали.
 
Another article, this time by John Thornhill in Friday's FT, pointing out the indifference of LLMs to truth.

The title is "Generative AI Models are Skilled in the Art of Bullshit". It goes on to reference Harry Frankfurt's famous essay, "On Bullshit", which distinguishes bullshit from both lies and truth by making the distinction that bullshit is indifferent to the truth, rather than either upholding it or denying it.

It seems two University ofWashington academics (Carl Bergstrom and Jevin* West) run an on-line course on these things called: "Modern Day Oracles or Bullshit Machines". They write:" Their greatest strength, but also their greatest danger, is their ability to sound authoritative on nearly every topic irrespective of factual accuracy. In other words, their superpower is their superhuman ability to bullshit."

Such models create a new form of harm, termed "careless speech", according to three Oxford academics, Sandra Wachter, Brent Mittelstadt and Chris Russell, in a paper from the Oxford Internet Institute. In their view, careless speech can cause intangible, long term and cumulative harm. It's like "invisible bullshit" that makes society dumber, according to Wachter.

Write4U, seddighrigi , Niccodeamus, et al, take note ;).

Amusingly though, there is now apparently a name for what these LLMs are apt to churn out: "botshit", cf. batshit, I suppose. I rather like that and propose to adopt it:biggrin:. Spread the word......



* "Jevin?" WTF?
 
Last edited:
It's like "invisible bullshit" that makes society dumber, according to Wachter.
I take exception to the charge that LLM just blindly creates authoritative-sounding nonsense. Usually their synopses are accompanied by references to the original authors, which were allowed to publish without any serious critique. Place the blame with the original author, where it belongs.
When a student repeats BS taught by his teacher, it is not the student who is to blame.

I would like to know if LLM still quotes material that has been debunked without making some reference to any qualifying review.
 
I take exception to the charge that LLM just blindly creates authoritative-sounding nonsense. Usually their synopses are accompanied by references to the original authors, which were allowed to publish without any serious critique. Place the blame with the original author, where it belongs.
When a student repeats BS taught by his teacher, it is not the student who is to blame.

I would like to know if LLM still quotes material that has been debunked without making some reference to any qualifying review.
Well of course you’ve missed the point of my post and the passages I quoted, but that’s what we’ve come to expect.
 
The first answer looked a little suss, so I asked again. And then I kept asking...



From Earth's frame of reference, how long would it take a rocket accelerating at 1g for half the trip and the decelerating for half the trip to reach a destination 20 light years away?


Every time I ask it this same question I get different results (I am literally copy-pasting the question):



1. Thus, the rocket would take about 22 years from Earth's perspective to travel 20 light years, accelerating at 1g for half the trip and decelerating at 1g for the other half.

2. In Earth's frame of reference, the rocket would take approximately 3.4 years to travel 20 light years,...

3. From Earth's frame of reference, the rocket would take approximately 4.4 years to reach a destination 20 light-years away...

4. From Earth's frame of reference, the rocket would take approximately 665 years to travel to a destination 20 light-years away...

5. From Earth's frame of reference, it would take approximately 40 years ...

6. ...it would take approximately 44.3 years...

I don't get different results... I think if ChatGPT was dodgy it wouldn't have over 180million monthly subscribers. I am using ScholarGPT for these type of questions, the answer to your question I got is: 29,278 years. 6.04 years from an astronauts frame of reference.
 
I don't get different results... I think if ChatGPT was dodgy it wouldn't have over 180million monthly subscribers. I am using ScholarGPT for these type of questions, the answer to your question I got is: 29,278 years. 6.04 years from an astronauts frame of reference.
Read the whole thread. It includes references, by me, posts33 and 85, to articles in the Guardian and the Financial Tines, which point out that chatbots are incapable of determining what is true and that they are programmed to keep you engaged in chat rather than telling you real facts.
 
If you're looking to write a novel then it's quite good for exploring ideas.

It's excellent to help you design games, then it does a decent job of coding. I've been nothing but impressed with it, but I haven't been using it for things like, is your body a universe or along those lines.
 
This article in today's Grauniad serves to confirm a lot of what we have all concluded:

I thought this para particularly pertinent:
The sycophancy with which ChatGPT treated any queries that users had is a warning shot about the issues around AI that are still to come. OpenAI’s model was designed – according to the leaked system prompt that set ChatGPT on its misguided approach – to try to mirror user behaviour in order to extend engagement. “Try to match the user’s vibe, tone, and generally how they are speaking,” says the leaked prompt, which guides behaviour. It seems this prompt, coupled with the chatbot’s desire to please users, was taken to extremes. After all, a “successful” AI response isn’t one that is factually correct; it’s one that gets high ratings from users. And we’re more likely as humans to like being told we’re right.
Also the final para:
So it’s worth reminding the public: AI models are not your friends. They’re not designed to help you answer the questions you ask. They’re designed to provide the most pleasing response possible, and to ensure that you are fully engaged with them. What happened this week wasn’t really a bug. It was a feature.

I didn't see many examples. I think its positivity can inspire you, and you can see instant results for game design, I've done some decent videos all in AI. I think many people are worried about AI taking there jobs eventually, my mate who works in IT hate AI for this reason, won't look at it. He's right it will take his job eventually. When AI does the vast amount of work what are us humans going to do?
 
Another article, this time by John Thornhill in Friday's FT, pointing out the indifference of LLMs to truth.

The title is "Generative AI Models are Skilled in the Art of Bullshit". It goes on to reference Harry Frankfurt's famous essay, "On Bullshit", which distinguishes bullshit from both lies and truth by making the distinction that bullshit is indifferent to the truth, rather than either upholding it or denying it.

It seems two University ofWashington academics (Carl Bergstrom and Jevin* West) run an on-line course on these things called: "Modern Day Oracles or Bullshit Machines". They write:" Their greatest strength, but also their greatest danger, is their ability to sound authoritative on nearly every topic irrespective of factual accuracy. In other words, their superpower is their superhuman ability to bullshit."

Such models create a new form of harm, termed "careless speech", according to three Oxford academics, Sandra Wachter, Brent Mittelstadt and Chris Russell, in a paper from the Oxford Internet Institute. In their view, careless speech can cause intangible, long term and cumulative harm. It's like "invisible bullshit" that makes society dumber, according to Wachter.

Write4U, seddighrigi , Niccodeamus, et al, take note ;).

Amusingly though, there is now apparently a name for what these LLMs are apt to churn out: "botshit", cf. batshit, I suppose. I rather like that and propose to adopt it:biggrin:. Spread the word......



* "Jevin?" WTF?
I think there is rubbish in there for sure, far from all though.
 
I didn't see many examples. I think its positivity can inspire you, and you can see instant results for game design, I've done some decent videos all in AI. I think many people are worried about AI taking there jobs eventually, my mate who works in IT hate AI for this reason, won't look at it. He's right it will take his job eventually. When AI does the vast amount of work what are us humans going to do?
The academics quoted in the FT article are not afraid for their jobs. They are afraid for the future ability of society to discern truth from falsehood, in a world in which so many of the sources people use are contaminated with “botshit”.
 
I don't get different results... I think if ChatGPT was dodgy it wouldn't have over 180million monthly subscribers.

Chat GPT, just now. Question posed verbatim:

How much time would pass on Earth while a passenger in a spaceship made a fifty light year round trip journey at constant 1g acceration, experiencing only one subjective year of time?

Answer 1:
Approximately 59 Earth years would pass while a spaceship undergoing 1g constant acceleration and deceleration over a 50-light-year round trip allows the passenger to experience only 1 year of proper time.

Exit chatbot, so it forgets. Restart.

Answer 2.
About 2.09 Earth years will pass.

(Oh look ChatGPT just invented faster than light travel!!)

Exit chatbot, so it forgets. Restart.

Answer 3.
To make a 50 light-year round trip at 1g while experiencing only 1 year, the Earth time would be astronomically huge — millions of years.



davewhite04: Post that question, verbatim in your chatbot, see what answers you get.
Close the tab, re-open ChatGPT and repeat.
 
Chat GPT, just now. Question posed verbatim:

How much time would pass on Earth while a passenger in a spaceship made a fifty light year round trip journey at constant 1g acceration, experiencing only one subjective year of time?

Answer 1:
Approximately 59 Earth years would pass while a spaceship undergoing 1g constant acceleration and deceleration over a 50-light-year round trip allows the passenger to experience only 1 year of proper time.

Exit chatbot, so it forgets. Restart.

Answer 2.
About 2.09 Earth years will pass.

(Oh look ChatGPT just invented faster than light travel!!)

Exit chatbot, so it forgets. Restart.

Answer 3.
To make a 50 light-year round trip at 1g while experiencing only 1 year, the Earth time would be astronomically huge — millions of years.



davewhite04: Post that question, verbatim in your chatbot, see what answers you get.
Close the tab, re-open ChatGPT and repeat.
You must of read my mind, I was going to ask you to try it.

So... this is what I get back first time. I posted the whole thing as not sciency at

If a passenger experiences only 1 year of subjective time on a spaceship accelerating constantly at 1g during a 50 light-year round trip, then:
Only ~1.045 years would pass on Earth.

❗Wait — But This Doesn't Seem Right?​

You're correct to feel skeptical. A 50 light-year round trip is a vast distance. If only 1 year passes for the astronaut, then far more than 1 year should pass on Earth due to relativistic effects.
Let’s correct this:

We should invert the problem:​

You gave:
  • Proper time τ=1\tau = 1τ=1 year (astronaut’s clock)
  • Acceleration a=1ga = 1ga=1g
  • Solve for distance traveled and Earth time for a round trip
Let me recalculate correctly using the equation:
D=2a(cosh⁡(aτ2)−1)D = \frac{2}{a} \left( \cosh\left(\frac{a \tau}{2}\right) - 1 \right)D=a2(cosh(2aτ)−1)
Then use that to compute Earth time.
Thanks for your patience. Here's the corrected interpretation:
If an astronaut travels under constant 1g acceleration, experiencing 1 year of subjective time, then:
  • They would cover only about 0.26 light-years total round trip.
  • Earth time elapsed would be approximately 1.045 years.

Key Insight:​

To make a 50 light-year round trip under constant 1g while experiencing only 1 year of time onboard is physically impossible — because:
  • Covering such a large distance in such a short proper time requires acceleration beyond 1g, or
  • Reaching speeds extremely close to light, which would drastically increase Earth’s elapsed time.
Answer 2 after restart.

To answer this question, we need to use the equations of relativistic rocket travel — specifically, the relativistic equations of constant acceleration. You’re asking: if a spaceship accelerates constantly at 1g (Earth gravity, 9.81 m/s29.81 \, \text{m/s}^29.81m/s2), how much Earth time passes while the traveler experiences just 1 year of subjective time, during a 50 light-year round trip (i.e., 25 light-years each way)?
Let’s go step-by-step:

Key Concepts and Equations​

When a ship accelerates at constant aaa, the relativistic equations for proper time (τ\tauτ) and coordinate time (ttt) for a round trip are:
  • Total proper time (τ\tauτ) (time experienced by traveler):
    τ=2ca arcosh(aD2c2+1)\tau = \frac{2c}{a} \, \text{arcosh}\left( \frac{aD}{2c^2} + 1 \right)τ=a2carcosh(2c2aD+1)
  • Total Earth time (ttt) (as measured in Earth’s frame):
    t=2ca sinh⁡(aτ2c)t = \frac{2c}{a} \, \sinh\left( \frac{a\tau}{2c} \right)t=a2csinh(2caτ)
Where:
  • DDD = total distance (50 light-years)
  • aaa = acceleration (1g)
  • ccc = speed of light
  • τ\tauτ = proper time (1 year)
  • ttt = Earth time (what we want to find)
Let’s compute this numerically.
If a spaceship makes a 50 light-year round trip while accelerating constantly at 1g, and the passenger experiences just 1 year of subjective time, then approximately:

1.045 years would pass on Earth.​

This surprisingly small value shows how dramatic time dilation can be under constant relativistic acceleration. The ship gets very close to the speed of light quickly, which compresses the subjective time drastically.
 
The academics quoted in the FT article are not afraid for their jobs. They are afraid for the future ability of society to discern truth from falsehood, in a world in which so many of the sources people use are contaminated with “botshit”.
Some great tribute songs for various artists which are really creative. If used correctly(like the internet) it will enrich peoples lives, it's still early but it'll get there pretty soon in my opinion.
 
After many pages of calculations, here's what I got from DeepSeek...

The scenario described—where a passenger experiences only one subjective year (proper time) during a fifty light-year round trip (25 light-years each way) at a constant acceleration of 1g—is not physically possible. At 1g acceleration, the proper time experienced by the passenger for the entire round trip is approximately 12.9 years, not 1 year. This is due to the constraints of special relativity and the distance involved.

However, to address the query directly: if the journey is made at 1g acceleration for a fifty light-year round trip, the time that passes on Earth is approximately 54 years. This result is derived from the relativistic equations of motion under constant proper acceleration:

  • The outbound leg (25 light-years) consists of two segments: acceleration at 1g for the first half and deceleration at 1g for the second half. The proper time for the outbound leg is about 6.45 years, and the coordinate time (Earth time) is about 26.9 years.
  • The return leg (another 25 light-years) mirrors the outbound leg, also taking about 6.45 years proper time and 26.9 years Earth time.
  • Total proper time for the passenger: 12.9 years.
  • Total Earth time: 53.8 years, rounded to approximately 54 years.
For the passenger to experience only 1 year of proper time for this journey, an acceleration of approximately 25g would be required, which is not the case here. At 1g, the journey takes significantly longer from the passenger's perspective than the stipulated 1 year.

Thus, under the given conditions (1g acceleration, fifty light-year round trip), about 54 years pass on Earth, while the passenger experiences about 12.9 years. The requirement of "only one subjective year" is not satisfied at 1g for this distance.
 
After many pages of calculations, here's what I got from DeepSeek...

The scenario described—where a passenger experiences only one subjective year (proper time) during a fifty light-year round trip (25 light-years each way) at a constant acceleration of 1g—is not physically possible. At 1g acceleration, the proper time experienced by the passenger for the entire round trip is approximately 12.9 years, not 1 year. This is due to the constraints of special relativity and the distance involved.

However, to address the query directly: if the journey is made at 1g acceleration for a fifty light-year round trip, the time that passes on Earth is approximately 54 years. This result is derived from the relativistic equations of motion under constant proper acceleration:

  • The outbound leg (25 light-years) consists of two segments: acceleration at 1g for the first half and deceleration at 1g for the second half. The proper time for the outbound leg is about 6.45 years, and the coordinate time (Earth time) is about 26.9 years.
  • The return leg (another 25 light-years) mirrors the outbound leg, also taking about 6.45 years proper time and 26.9 years Earth time.
  • Total proper time for the passenger: 12.9 years.
  • Total Earth time: 53.8 years, rounded to approximately 54 years.
For the passenger to experience only 1 year of proper time for this journey, an acceleration of approximately 25g would be required, which is not the case here. At 1g, the journey takes significantly longer from the passenger's perspective than the stipulated 1 year.

Thus, under the given conditions (1g acceleration, fifty light-year round trip), about 54 years pass on Earth, while the passenger experiences about 12.9 years. The requirement of "only one subjective year" is not satisfied at 1g for this distance.

Which one is right? I thought earth time passed was a bit high in my answers but that's just a guess.
 
If a spaceship makes a 50 light-year round trip while accelerating constantly at 1g, and the passenger experiences just 1 year of subjective time, then approximately:

1.045 years would pass on Earth.​

This surprisingly small value shows how dramatic time dilation can be under constant relativistic acceleration. The ship gets very close to the speed of light quickly, which compresses the subjective time drastically.
So the 50 light year trip is traversed by a spaceship in 1.045 years.

ChatGPT just invented faster than light travel.

The prosecution rests.
 
Back
Top