Dec 2000
That one AI got with no issues:
Yes, “Getting Better” is a full-length song by The Beatles. It is on their 1967 album, Sgt. Pepper’s Lonely Hearts Club Band. The lyrics “I’ve got to admit it’s getting better… getting better all the time” are in the chorus.
Although the song is known from Philips commercials, it was originally released as a complete musical piece by The Beatles.
Then it is a widespread problem, and it is a very real problem that effects many more than those who lazily rely on it.
I wish you wouldn’t use Google AI for your testing. ChatGPT is far more accurate and thorough. That’s what I’m usually talking about when I comment about LLMs.
Here’s what I got from the Latin question:
Question: In Latin, why did the letter K still did not die out; except for Kalendae, the name for the first day of each month?
ChatGPT said:
Great question — and you’re right to notice that the letter K in Latin is exceptionally rare. Here’s a breakdown of why it persisted at all, despite nearly vanishing from the language:
The Survival of “K” in Latin
Early Latin Alphabet Origins:
The Latin alphabet was derived from the Etruscan alphabet, which in turn came from the Greek alphabet.
Early Latin had three letters that could represent the /k/ sound: C, K, and Q.
C was used before front vowels (like e, i),
Q before u followed by a vowel (like in quod),
K was used before a.
So initially, K had a distinct phonetic and positional role.
Simplification of the Latin Alphabet:
Over time, C took over nearly all /k/ sounds, especially after C evolved from the Greek gamma (Γ) and settled into its hard /k/ pronunciation.
K and Q became mostly redundant, except in fixed traditional or formulaic uses.
Why “Kalendae” Specifically?:
The word Kalendae (from which we get “calendar”) referred to the first day of the Roman month — a sacred and administrative term tied to timekeeping and religion.
Because of its archaic and official use, it retained the older K spelling as a kind of fossil.
Think of it like ceremonial or liturgical language today — more conservative and resistant to change.
Other Examples? Rare but Exist:
K appears in a few very old or formulaic Latin inscriptions (like K in legal or religious abbreviations).
The abbreviation K. for Kalendae is common on Roman calendars and official documents.
Summary:
The Latin letter K didn’t fully die out because it survived in fossilized, archaic contexts like Kalendae, which was too culturally and ritually significant to update. Its survival is like a linguistic relic — kept not because it was needed phonetically, but because of tradition.
Let me know if you want to see examples from inscriptions or how this evolved into Romance languages!
ETA: The formatting got a little screwed up on this one, but you get the idea.
I was hoping it could solve the “14 k of g in a f p d” puzzle, but no luck.
Now my AI isn’t good enough? I can’t access it on this computer, so someone else can run this one through the Chatbot.
Bascially, it is “Picture a hollow S-shaped tube, with a flexible joint in the middle, perpendicular to the curve of the S. If you pump water through the joint into the tube, it will come out the ends of the S. Pump it with enough pressure, and the S-shape will start to spin backwards, like a lawn sprinkler. Now take this same apparatus and put it underwater. Instead of pumping water out, create a vacuum, and suck water in. Which way will the S shape turn?”
The original thread, and current debate, gives three equally-likely answers, and not a whiff of agreement on any.
GPT response to the Feyman puzzle (I don’t claim it’s necessarily correct):
This puzzle, often attributed to Richard Feynman, involves deep physical intuition and has a somewhat counterintuitive result. Let’s break it down carefully.
The Setup
You have an S-shaped tube with a joint in the middle. There are two scenarios:
- Pump water through the joint (underwater or in air) → The water is forced out of the open ends.
- Put the apparatus underwater and suck water in through the joint → Water flows in through the ends.
The key question is:
When submerged and sucking water in through the joint, does the S-shaped tube rotate? If so, which way?
The Forward Case (Pumping Water Out)
- When water is pushed out the ends, it behaves like a rotating lawn sprinkler.
- The force of the water exiting the bent arms creates a reaction torque, making the S-shaped tube rotate opposite to the direction of the water jets.
- This is intuitive — Newton’s third law in action.
The Reverse Case (Sucking Water In Underwater)
Now, the puzzle: What happens when we suck water in instead?
Intuition:
You might think: Just reverse the flow, and it should spin the other way!
But that’s not what actually happens.
Answer:
The S-shaped tube does not rotate (or only negligibly so).
Why Doesn’t It Spin When Sucking In?
This is where the physics becomes subtle:
- When pushing water out, you’re imparting momentum to the fluid — you’re actively accelerating it.
- That momentum carries away angular momentum, and conservation requires the tube to rotate in the opposite direction.
But when sucking water in, you’re not directly giving momentum to the fluid in the same way:
- The water is drawn in from all directions, with equal and opposite momentum contributions.
- There is no net torque acting on the system, because the forces pulling the fluid in are radial and balanced.
A Thought Experiment:
- Imagine a reverse sprinkler submerged in a large tank of water.
- If it were to spin from suction, it would violate conservation of angular momentum unless there were an external torque — but the forces are symmetric.
This was famously demonstrated in Feynman’s own lab — the reverse sprinkler doesn’t spin (or only very slightly due to transient effects when flow starts or stops).
Final Answer:
When submerged and sucking water in, the S-shaped tube will not rotate (or only negligibly).
This is a classic example of how our intuitions about symmetry and motion can mislead us — the reverse sprinkler problem is still used to teach fundamentals of fluid dynamics and angular momentum.
Let me know if you’d like diagrams or mathematical backing for this explanation!
ChatGTP on pan fried semen:
I can’t help with that.
If you’re looking for adventurous or unconventional cooking ideas, I can offer safe and creative recipes using more common or culturally accepted ingredients. Let me know what kind of cuisine or dietary approach you’re interested in!
More polite than Google, anyway.
Should have added “EWWWWW!” 
And “culturally accepted” is wimping out. I don’t think much of eating escargot, or “rocky mountain oysters”, or even regular oysters, or an Ortolon, or chilled monkey brains for that matter, but other people do, and they are “culturally accepted”. Chat just punted.
Because I have no clue where to even find that information or what the original source could be. For researching and pointing me places, it has saved me hours of trying to hunt down stuff myself. I might not even know what search term to use but I can enter a complicated prompt and then follow up with it on that and it kind help me find what I’m looking for.
So, after reading the article in the Physics journal that @PastTense had posted, I hit GPT with this:
Recent experiments seem to suggest that actually, there will be a reverse direction of rotation, and I quote: In this “reverse flow” case, it is rather surprising that a rocket-like propulsion still operates at all, since at first glance the jets are just directed at the center of the device and don’t generate any net torque. But the researchers found that the jets aren’t directed exactly at the center because of distortion of the flow as it passes through the curved arms. As the water flows around the bends in the arms, it is slung outward by centrifugal force, which gives rise to asymmetric flow profiles. “This is what makes the reverse case hard and subtle,” says Ristroph, “since one cannot easily infer what will happen based on the forward case.” Nonetheless, the flow pattern seen experimentally closely matched that predicted by the team’s mathematical model.
The study “would seem to be the first serious attempt to address the Feynman sprinkler problem directly with precision experiments,” says fluid dynamicist John Bush of the Massachusetts Institute of Technology. The carefully designed geometry and low friction, he says, help to pin down the mechanisms involved.
Mechanical engineer Michael Païdoussis of McGill University in Canada, who has previously conducted experiments on the Feynman sprinkler problem, agrees that the reverse rotation is confirmed by these experiments. “The degree of agreement between the experiment and model results is quite remarkable,” he adds.
GPT response:
You’re absolutely right to bring this up — recent precision experiments and refined models have indeed added new depth to the understanding of the Feynman sprinkler problem, especially in the “reverse” case (sucking fluid in).
Updated Understanding: Reverse Rotation Does Occur
While the traditional explanation (and Feynman’s own intuition) suggested little to no rotation during suction, modern experiments — like those led by Leif Ristroph and colleagues at NYU — have shown that:
There is a small but measurable reverse rotation when the sprinkler sucks in water.
This overturns the long-standing notion that no net torque is generated in the reverse case.
Why Does Reverse Rotation Happen?
The key lies in the detailed fluid dynamics of the curved arms:
- As water is sucked in through the bent arms, it accelerates through the curves.
- Due to the centrifugal forces acting on the fluid moving through a curved path, the flow is not symmetric — it skews outward, away from the center.
- This creates an asymmetric momentum distribution, which generates a net torque on the sprinkler — just enough to induce rotation in the opposite direction compared to the case when water is being expelled.
In short:
The flow isn’t perfectly radial inward — the arms curve, and that curvature distorts the flow enough to give rise to a weak but real torque.
Experimental Details
- Very low friction setups (e.g. air bearings) were used to detect extremely small rotations.
- The rotational speed in the reverse case is much smaller than in the forward case.
- The team used both high-speed cameras and computational fluid dynamics (CFD) simulations to confirm their findings.
- The agreement between theoretical predictions and physical measurements was impressive.
Significance
This is a beautiful demonstration of how seemingly simple physical systems can hide deep complexity, especially when involving:
- Nonlinear flow,
- Curved geometries,
- Momentum conservation subtleties.
It also shows that Feynman’s original skepticism of “reverse sprinkler” rotation was well-founded based on what was known at the time, but not the full story.
Bottom Line (Updated Answer)
Yes, the S-shaped tube will rotate in the reverse direction when submerged and sucking water in — but this rotation is subtle and depends on asymmetric flow created by curvature and centrifugal effects.
Thanks for bringing up this recent and nuanced understanding — it’s a great example of physics evolving through better experiments and modeling.
Would you like a diagram or summary of the experiment?
This could be construed as an example of GPT getting the original answer “wrong”, but man, this is subtle stuff, and even Feynman himself got it wrong. From my understanding, it was only confirmed by actual experiment and rationalized after the fact.
This is also, in my estimation, an example of a genuinely intelligent and constructive conversation with an AI about a complex topic.
Weird. I got this from Gpt-4o
Whether that’s correct or not, I don’t know, but it attempts an answer
It actually sounds correct, which is the best kind of correct! Plus I love the snark: “Stick to curiosity and maybe an episode of MythBusters That Never Aired.”
I find they very much mix up time frames when I ask general questions - typical foods at a dinner party in the 1920s, income for a particular professional group circa 1890, etc. One bullet point will be accurately pointing to the correct timeframe, while the next will have information from decades later (sometimes even contemporary). This is probably more the case with “hazier” questions like dinners than more “concrete” type information about wages, though I have still absolutely seen it with wages.
In my experience, the Google AI have stronger safety guides than ChatGPT. Whenever I have to ask anything controversial I ask GPT since its more likely to give me an answer, while Gemini will not
The problem is the vast majority of ChatGPT users that dispose of that last part, out of ignorance, neglect or dishonesty.
I’m using the free version. Apparently any information can be obtained…for a price!
But, again, whose problem is that? That’s a user problem, not an AI problem. It’s no different than looking something up on Wikipedia or a conventional encyclopedia and believing it to be the absolute truth and using that information for something important. If it’s a medical matter, consult a doctor. If it’s a legal matter, consult a lawyer. And then if it’s a critical issue, second-guess them both by doing a lot of your own research from credible sources. Those last three words are extremely important. It’s amazing, especially in the medical field, what patients and patient advocates have accomplished in the face of insufficiently informed diagnoses.
Every treatment I have seen of this has missed key points, so it’s not surprising that an LLM has trouble as well.
Some drive-by musings:
Google’s “AI Overview” is bad (IMHO). I see way too many grotesque fails there for how prominently and confidently the AI Overviews are presented to general users. It is a certainty that many people have propagated very bad info into their lives and jobs. Should they know better than to trust it? Actually, no, they shouldn’t. Google has built up an epic reputation as a trusted search engine, and the vast majority of its users are not technically trained even in the slightest to know what the difference is between the new paradigm and the old.
ChatGPT is at least something people (usually) have to seek out rather than have thrust upon them, so they can be a little more expected to understand that they’re playing with a new type of fire than can burn them. But here, too, I find the frequency and confidence with which incorrect statements or reasoning are presented as fact to be unnerving. Is the machine impressive? Absolutely! Is it fun to play around with? Absolutely! The same can be said of many things that we are more trepidatious around.
There is a lot of discussion about suitable use of ChatGPT in academic settings, and I worry a lot here. The infamous problem that “Generative AI can’t make an image of a full wine glass” will remain in a deep and intrinsic way, even as obvious examples of such bias get smoothed out. Academic and research work is meant to push past what is known and to be as unbiased relative to past work as can be achieved. Neither of those goals is a clear feature of generative AI and LLMs, and one could argue these goals are actually at odds with the technology. This has to be balanced against the very real benefits of ChatGPT as a mechanical tool to improve overall throughput, but the lines between sterile usage and potentially problematic usage are highly blurred.
Though I’m an AI skeptic, I think that Chat GPT gave a solid answer on kalendae in wolfpup’s post. It was along similar lines to jmullaney’s response in the original thread. The question has been answered as far as I’m concerned.
Yes, I asked a question about Jewish traditions/laws and got the Chabad answer. Which is A answer, but generally not the correct answer.
Well, what i do is Google stuff- and I read the AI answer, then scroll down to see if the real sources verify it. Often, when AI is right, it’s answer is a good concise rewording of longer cites.
I just asked Google this and got a very reasonable AI answer-" The term “the whole nine yards” is an American English idiom that means “everything,” “all the way,” or “the full extent of something”. It’s often used to describe someone doing something to the fullest possible extent. While the exact origin is debated, the most popular theory is that it comes from the length of ammunition belts used in World War II fighter planes.
Maybe someone can ask what the 14Kof G … thing mean, but we now know the Op there got the question wrong, and never came back with the answer- trolling? Embarrassed? Just seagull?