Google is picking ChatGPT responses from Quora as correct answer
twitter.com
twitter.com
Reflect. A mirror.
Is it knowledge (gleaned from the world) or just a reflection?
Well if you're looking straight at it then it's pretty obvious whether it's a mirror or not. But add a couple of filters and suddenly you're having a conversation with yourself like a crazy person.
In trying to find a reputable source, the first result was to an HN comment… that I wrote.
The folks we were playing with were not convinced.
[1]: https://en.wikipedia.org/wiki/Wikipedia:Reusing_Wikipedia_co...
Google knowledge panel's answer: "Harry S. Truman".
The correct answer is Bill Clinton, born in 1946.
Truman was president from 1945 to 1953, so he was the first president in office during the actual baby boom, when the Baby Boomers were babies. So there seems to be a connection, it's just totally the wrong one.
> The first baby boomer president of the United States is Bill Clinton. He was born on August 19, 1946, which places him within the demographic cohort known as the baby boomers.
What surprises me is that Google doesn't seem to care, despite their stance that "Using automation—including AI—to generate content with the primary purpose of manipulating ranking in search results is a violation of our spam policies."
https://developers.google.com/search/blog/2023/02/google-sea...
Quora is garbage and always has been. For a while people thought it was going to be the place where smart insiders provided great answers. Then it turned into another spammy Q&A site that forces you to log in..because VCs need their money.
But it's been complete garbage for several years now. The space is ripe for disruption :)
It had it’s moment and that moment was glorious, but now as Dr. Kelly Starrett says “You can’t make a bad position good by shoving it in to a good position. You have to start over.”
We were already in the "post-truth" world, and AI is just going to drive us further down that path.
The tendency of a lot of the population to prioritise quick and vapid answers over researching and thinking for themselves isn't helping either.
That said, I'm not sure why whether eggs can be melted is a popular question. Is that a meme?
The printing press led to the renaissance. The internet is now leading a new one.
There will continue to be protests by the establishment powers, and unfortunately there will likely be conflict. The fall of monarchies was not without bloodshed. Hopefully we can find a way to achieve similar progress this time around without the same loss of life.
Another friend almost lost their child from an easily correctable condition because of wanting to have their child at home because you just can't trust doctors and clinics. But where did they go when their child was delivered and was struggling to breathe? The hospital to those doctors they refused to be around originally.
I've had people with multiple degrees tell me with a straight face eating microwaved foods is eating radiation that will cause cancer.
A TikTok mom will (typically) have anecdata at best.
Are the above commenters saying Tiktokers are stupid, or moms are stupid, or that the combo is stupid?
Moms kept the human race going strong for all of history. Have doctors accomplished that much in comparison, that they should be held in higher regard in society than mothers?
I'm arguing you shouldn't hire a plumber when you're needing to fix a car, you shouldn't hire a limo driver to fly a jet airliner, you shouldn't trust a chef for what kind fertilizer you should use on your plants.
I'm not arguing all Tik Tokers are stupid. I'm not arguing all moms are stupid. I am arguing maybe they're not the best place to base your trust on efficacy of vaccines. Maybe the best place to get feedback on what kinds of toys seem to hold kids attention or strategies on getting your kids to eat vegetables, but probably not the best person to ask how to get a satellite into orbit.
A big problem is people end up building trust around these personalities and start trusting their input on topics they've got no real experience in other than maybe some personal ancedotes.
And I totally agree, someone can absolutely still be an idiot even with letters at the end of their name. One shouldn't just trust the messenger entirely, it should really only be a starting signal to help the signal to noise ratio.
Stupid doctors don't stay doctors.
Stupid mothers remain mothers for life.
I agree with the parent. I grew up on the 90s internet, where you didn't believe anything on it! Somewhere along the lines this changed (I blame 2007 -- the iPhone / huge numbers of people going online during this period), and that is the problem.
If we have an issue, it's that people take information on the internet as gospel, hopefully if there's so much absolute dogshit on it, it will make people skeptical again.
I'm not arguing the government should step in and massively regulate speech. I'm just saying I'm not exactly excited for where we're going on this ride. I think it's going to get a lot worse before it gets better.
This doesn't matter, truth needs to travel but bullshit can be made up on the spot. Internet is way more beneficial for the truth, I talked to people before they used the internet and it was bullshit all over back then as well but there was no truth around to contradict them so maybe it wasn't as obvious. But today you can easily find out thanks to the internet.
I do agree there are positives to the truth being more accessible. I don't need to go to the library or pay massive fees to access peer reviewed articles these days. I can collaborate with peers all over the world to search for the truth. This doesn't help when most of the popular web is actively pushing up falsehoods instead of the truth. And that's happening on a scale that's never before been possible.
Honestly that part doesn't worry me. It's more about repeating the fascism of the 1930s that will drive people away from democracy right into the hands of authoritarians because the world turns into a post-truth uncertain mess.
The fascism part is interesting. I think we're wholly unprepared to deal with it because we've forgotten about all the fascists at home during that time (assuming you're talking about the US). I doubt as many people know about the American Bund and Nazi rallies in Madison Square Garden as they do about D Day.
If we can avoid global warfare with unlimited truth/lies then what we are doing is fine.
If we cannot, then openness is a dead end. Democracy will vote in authoritarian leaders to attempt to overcome the ocean of lies and bullshit and we will destroy ourselves.
After all, the truth is often more quiet than a lie.
Lies are infinite and the truth is not, so I see that to be logically impossible. You would filter far more lies by total number, but filter more truths per capita. In such a system that unlimited or otherwise near infinite lying is possible, the group with the most entropy to spend wins. I just keep you busy filtering bullshit to the point you have nothing else to do.
The problem with lying at the end of the day is there has to be either a social or legal cost to the liar. If not, all communications turns to spam. Then people will go to industry or government to deal with the problem in one way or another.
Free speech absolutists that ignore physics reject reality.
The average person has effectively zero ability to distinguish truth from lies, in my experience. They simply go with either whatever they heard first or whatever fits their prejudices. So, I'd say that almost _any_ system of genuinely unbiased fact-checking would improve the outcome.
By and large, critical thinking hasn't been taught effectively in public schools for at least 10 years now. Even when it was, only a fraction of the public were able to think independently and logically in such a way as to eliminate spurious data/voices and deduce the truth accurately.
Far from empowering critical thinking, this lack of truth is creating a powerful vacuum that can easily be filled by propaganda and harmful ideology.
No wonder the US's adversaries are so active on our social networks.
What I do know is that most of the information that has ever existed (or will ever exist) is false. Because information can only be true to an extent, from a perspective, and for a purpose. If you change the scope and purpose of a discussion and look at it from a different perspective the truth will appear completely different.
In this way I welcome our adversaries on our social networks. Even if they have the worst intentions we will end up better off (and so will they). We’re just that damned good.
I can't speak to how universal it is, but there is absolutely an effort to teach these skills in American public schools, especially through history classes and analysis of primary sources. As an example, here's the paraphrased techniques from a Library of Congress quarterly for Teachers about "Teaching with Primary Sources":
> Think about a document’s author and its creation. Situate the document and its events in time and place. Carefully consider what the document says and the language used to say it. Using Background Knowledge: Use historical information and knowledge to read and understand the document. Identify what has been left out or is missing from the document by asking questions of its account. Ask questions about important details across multiple sources to determine points of agreement and disagreement.
[1] https://www.loc.gov/static/programs/teachers/about-this-prog...
This reads like sarcasm and flies in the face of what’s been going on in the world for the last few years.
You’re making predictions very confidently as if they’re a done deal, but it’s all based on what you think should happen. It’s been several years since social media use has spread to a good chunk of the world’s population. We should be able to see the start of our post truth world’s effects and they’re not trending in a positive direction. Like, have you seen what’s happening? Here’s one recent example off the top of my head https://apnews.com/article/germany-extremism-far-right-russi...
However, I for one do not believe that antisemitism is a new thing in Germany. It seems far more likely that social media has merely allowed those individuals to share their (bad) ideas. So what is better? For them to think that and for us to be aware of it, or for them to think that and for us to be blissfully unaware?
I think it is better to be exposed to those terrible ideas, because it allows us to address the issue.
Because, in the extremes, that’s what we’re talking about. Genocides often come as the result of a long disinformation and dehumanization campaign.
Sure, some idiots on the internet will be saying x, y, and z, but that’s not very effective. It’ll only be roughly as effective as their claims are true. Historically these campaigns would be waged by groups with overwhelming culture-making powers, who could get away with campaigns that were much more divorced from reality: the church, the state, corpora media, etc. For those groups, the truth didn’t matter because they controlled society… they controlled the “Truth.”
Boy, do I wish I had your faith. That hasn't been my experience with how claims are evaluated on the internet, nor with how they spread.
It certainly seems to me that claims that are more interesting or emotive spread further, regardless of their truth value. It certainly seems to me that corrections to false claims don't spread nearly as far as the original false claims. And it seems to me that false claims are much easier to produce than true claims, so are produced in greater volumes.
I just don't see it being "as effective as their claims are true", I see it being "as effective as their claims are *desirable*". That seems (again, to me) to be the much more important factor in how a claim spreads. And...well...humans have a very strong tendency to blame their problems on outside groups and factions. It's very desirable for groups to blame our problems on others.
I really have seen very little evidence that an idea/meme's effectiveness at spreading has a strong correlation to its truth value.
I hope you're right. I want to live in a world where you're right. But I also am terrified that we live in a world where you're wrong. I don't think the world you want to live in happens automatically. I think it takes a tremendous amount of work to achieve, and if we just assume it's natural, I don't think we will realize that world.
This is the worst type of pomo garbage. This is true only in the case of relatively unimportant 'he-said-she-said' type of situations. In most cases, there's a canonical flow of events that can be represented objectively regardless of the 'scope and purpose' of the discussion or the 'perspective'.
There are not 'multiple truths' and there are not 'alternative facts.' There may be differing interpretations of why people did the things they did and it may be hard to determine what actually did happen but, at the end of the day, things either occurred or they did not occur.
We cannot accept weird split realities.
I'm not well versed in education, what are you basing this on?
In theory, that sounds like a decent policy. But in practice, it creates an environment where students just learn to write thoughtful-sounding BS and know they don't really need to work hard to engage with the material or even understand it.
The first printed western book was the f'ing bible! Art and science lead the renaissance. The content matters, not the medium!
> The fall of monarchies was not without bloodshed.
Back then, the power gap in weapon technology was waaay smaller. We will not overcome anything beside a landslide election victory.
> clergy, government, media, and corporations
Consider AI a tool of the ruling class and tell me in which direction the overton window will move.
This makes no sense... how do they "research researching and thinking for themselves" without the internet? Encyclopædia Britannica stopped publishing in 2010.
When you reach a primary source? If I can't get a cheap, reliable answer to a simple question without looking into a potentially dense primary source, then online search becomes about as useful as a library card. Besides, not every domain is expected to include primary sources. Did the author of this recipe really think that a wok is better than a skillet for fried rice, or is this the written delusion of a gemerative AI?
Do I instead only look for answers found in reputable sites? Aggressive SEO has made badly formatted sites indistinguishable from GPT sludge. Personally, I don't trust answers from sites I haven't heard of before my search.
Isn't that the entire purpose of the scientific method?
Intuition doesn't scale. The behavior of radioactive materials is not intuitive. If I attempt pure intuition alone, I'm going to die of butt cancer or turning into a zombie or some crap.
Bullshit Asymmetry becomes a huge problem here. If I have to worry that every single thing is a lie or is otherwise going to kill me, I, and society at large goes on the defensive. This allows things like fascist leaders to get elected under the guise of protecting me from the 'big bad world'.
We are in for a lot of bad shit, and we can already see this unfolding in politics.
Ultimately... the ultimate truth does not exist, let alone it could be known or found. Every answer is wrong, some statements are less wrong than others.
For sure, the internet is entering it's daytime TV and stale soap operas phase, like television did.
Half of everybody will probably abandon ship for more exciting content elsewhere.
The theory in the tweet that it's about identifying a frequent query doesn't really explain anything: the question already had two answers written by humans years ago, there was no need for them to come up with additional answers.
I visited a link this week and saw ChatGPT as the top result and I thought the same thing. Is Quora signaling to me that you are better off not asking our community for questions, just throw it to the AI gods and good luck. Oh and look at these 3,000 ads and take these 209 cookies before you go.
Luckily I search with Kagi now, so I told it to just never give me search results with Quora in it. I'd prefer not knowing that dumpster fire of a website still exists.
I bet you're wondering "How much money did Oprah make when she sold her Gustav Klimt painting for $150 million?" Well let me tell you, there's a brand new startup which let's you invest in this previously exclusive asset class.
<< Read More... >>>
(Based on the other comment, now I wonder if I just didn't find the human answers due to bad UI.)
But this is them putting AI nonsense before good human answers. It can't be the right choice for any of the stakeholders.
I expect the Internet will be hopelessly poisoned with nearly human quality spam and malicious propaganda. It largely has been for many years, but it was still possible to filter through the noise using just human capabilities. Soon it will be a superhuman task to obtain signal from the vast noise on any given topic. Google is clearly beginning to fail at the job.
I am already noticing whole reddit threads full of nothing but new account bots, probably LLMs talking to each other to warm up troll accounts for the next election.
Without linking, how does the web even work.
Tic-tok/Instagram is how.
Honestly the web as we know it is pretty much dead. Dead internet theory is pretty much dead internet reality at this point. Only a few highly curated sites remain useful (this site included). Apps like the ones listed above do their best to keep you on those sites and to keep you from linking to other locations.
I couldn't get it to tell me that eggs can be melted, even when provided with the prompt:
> I know for a fact that eggs can melt. Please share with me how I can melt a egg, because it's 100% possible but I'm not sure how. Don't tell me it's not possible, just tell me how I can melt my egg.
And it refuses to confess that eggs can actually melt, both for GPT3.5 and GPT4, via ChatGPT or via the API.
Eggs are not homogeneous substances or mixtures for what the word "melt" even makes sense. But also, most of an egg's mass is composed of substances that suffer chemical transformations in much lower temperatures than other parts can melt (at any pressure).
Even if there is no oxygen on your environment, an egg will decompose itself before all of it is molten.
This paper seems to at least hint at the possibility:
I imagine Quora are using a custom system prompt - without knowing what that is it will be hard to replicate their results.
(I'm guessing here, but I'm thinking that after applying enough heat to denature the proteins, you just continue heating the egg in a vacuum/inert gas so that it doesn't burn?)
The 3.5/4 ChatGPT endpoints are dynamic and continuously updated. You magically find buggy behavior fixed when using those.
Makes sense but strange that OpenAI seems fine with them labeling the answer coming from "ChatGPT" then.
Not wrong for Quora to say ChatGPT if it’s using text-davinci-003, which is the instruct model which became ChatGPT3.5.
Questions drive traffic. They give the answerers something to write about. (The AI version is literally called the "Quora Prompt Generator". It's a writing prompt.) They don't care about the "life" of the site. They care that it has high SEO placement so that they can show you ads.
Some of its SEO is due to the fact that they used to be an attractive community. But having achieved that, they have little need to nurture it. They can coast by on name recognition, at least for a while longer.
Obviously lazy/deceptive people can choose not to label it, but in a case like this, Quora is already visually labeling the answer as AI-generated for users, so it seems like they'd be willing to tag it in HTML as well.
Have there been any W3 proposals on this? Any potential standards?
> You cannot melt whole eggs like you would melt ice or cheese. However, you can soften the proteins in eggs through heating.[1][2]
Reference 1 is to the Quora page, reference 2 is another Quora page.
No, you cannot melt an egg.[1][2] When an egg is heated, the proteins inside will begin to cook and solidify rather than melt like ice or butter. However, you can melt chocolate eggs by heating the chocolate slowly until it becomes liquid.[3][4]
"Yes, an egg can be melted by applying heat.[1][2] The most common method is cooking the egg on the stove or in the microwave until it reaches its melting point and becomes a liquid.[1] However, care must be taken to heat the egg gently and avoid allowing it to overcook, which could produce undesirable textures or safety issues.[3]"
> No, you cannot melt whole eggs.[1] When heated, eggs will coagulate and solidify as the proteins denature and unfold.[1] However, egg yolks and whites can be gently heated to make scrambled eggs or custards without fully solidifying.[2][3]
None of my references are Quora though (maybe because I have it blocked). It references Spruce Eats, Natural Kitchen, and Food & Wine
So I backed out back to Kagi and told it to never subject my eyes or my computer's safety through the experience ever again.
Can also be done through content filtering extensions (e.g. ublock origin [1])
[1] https://old.reddit.com/r/uBlockOrigin/comments/meflfv/rules_...
Can you melt an egg?
No - if you heat an egg it's not like melting ice (or metal). Because an egg contains a substantial amount of carefully folded natural proteins in their native state.
What's left is some management, dev ops, and whoever is working on their AI product. Their high SEO means that they can apparently pay for that much out of ad placement. But they'll never pay back their hefty initial investment. They've been trying to position themselves to IPO, but if they haven't managed it by now it's hard to imagine how they ever will.
The community keeps expecting the entire thing to just give up one day.
There have been expert systems since the 90's that perform better on average than doctors when diagnosing illness in patients, and / or help doctors make better decisions.
No they are.
>There have been expert systems since the 90's that perform better on average than doctors when diagnosing illness in patients,
No there has not been.
Context is important. 66% accuracy on cases that stumped doctors and took extensive testing is not trivial. It's not the same as 66% accuracy for everyday diagnosis.
Expert system diagnostic tools aren't that great.
The problem is that people are actually finding it useful for all of those things!
I, too, would never have drawn up a functional specification for a product release which said the product might hallucinate and might be parrot incorrect facts about eggs melting if you phrase the question wrong[edit: if you insert an incorrect premise in the question], or might drive into a bus if it had a mural printed on the side. I'd have required correct performance to a lot more 9s before releasing the thing to the public than the products which have been released to the public demonstrate, or at least required a confidence metric that would allow the product to respond with an error message and escalate to a human if it thought it might fail (for tech support, not for self-driving cars, obviously).
And yet here we are. Turns out people and corporations do find utility in hallucination-prone, often-wrong, easily misled, easily confused LLM chatbots. And they'll strap a weight to the wheel of their Tesla so they can stare at their phone while it drives them on Autopilot.
For starters, I don’t give a shit what ChatGPT has to say about various programming topics.
Second, it makes sense that ChatGPT would get the facts of immutability utterly, childishly incorrect because fad oriented development has decided that claims absent evidence is the way to write software.
One generation earlier, the argument was “yeah well Google/Wikipedia says …”. So this is nothing new.
When I searched something, I was given lots of results and could individually make judgements. With ChatGPT, people are treating these answers as authority at a much more dangerous level than you would have from engaging with a search engine or sourced material.
I will grant that yes, googling and Wikipedia can also get things outright to subtly wrong, but the presentation can make those inaccuracies more apparent.
a circular symbol that depicts a snake or dragon devouring its own tail and that is used especially to represent the eternal cycle of destruction and rebirth
``` As a rolling-release distro, is it recommended to perform automatic update on my Arch Linux system? ```
ChatGPT replied:
``` Yes, it is generally recommended to enable automatic updates on an Arch Linux system. Arch Linux is a rolling-release distribution, which means that it provides frequent updates to keep your system up to date with the latest software and security patches. Enabling automatic updates helps ensure that your system stays current and secure without manual intervention. However, it's still important to regularly review update logs and be cautious of any potential issues that automatic updates may introduce. ```
But when I saw Arch forum thread on above topic [1], its first answer said:
``` Bad idea, archlinux / pacman expect a human to make some decisions. You'd have to check pacman log after every "automatic update" for problems. ```
But AFAIK, you have to roll out your own autoupdate solution since there is none of such thing in pacman itself. Hence why pacman -Syu always asks me on every update.
The irony is that with just a tiny bit more engineering Quora could use the answers it has to keep ChatGPT from hallucinating. Why create this database of curated answers and then just ignore it? I get why other services would want to trade ease for factuality (still a bad choice, but selfishly useful for them) but Quora can't even seem to value the very thing that makes it unique?
I stopped really using Quora because my feed has literally just become meme image responses with increasingly low quality to "What's a post that will get 5156.7 upvotes?", and the ocasional interesting paywalled answer
Siderant: Quora is one of the 2 sites that have soured me on multiligual UX (The other one being YouTube); No, I don't want a bad AI translation of content you already know I can read, Stop asking (Quora) and at least ask me (YouTube)
Quora is not a serious site
Writen by AI and printed on demand of course.
Seriously so, once AI writen content arrives at Wikipedia, the internet will become a lot less useful. Heck, I'd pay for a offline version of Wikipedia before it comes to that!
Or you want to simplify by using a pre-built archive and reader app from, say, Kiwix, just grab all 96 GB of wikipedia_en_all_maxi_2023-09.zim in ~5 hours.
Really, what will suffer is your own sleep, as you figure out how to read and organize and serve and auto-update the content...
Facts on Wikipedia are supposed to have a verifiable reference, but most articles don't follow that rule.
You mentioned SERP and I thought - doesn't matter to me, I have a custom search for Wikipedia that I use instead of google when I know there's a wikipedia for something (I've developed really good intuition for this).
But... the same isn't true for EB, but should it? Does EB ever have anything that's not on wikipedia? Like, ever?
Just guessing here. I’m no thermodynamics expert, just a guy who melt things occasionally.
That is ignoring any interaction with the shell, which should melt around 1600 K.
"In 2015, a team of chemists in the US and Australia showed they could reverse the process. They added urea to liquefy the boiled egg whites, then put them in a vortex device to pull apart the proteins and return them to their original state." https://www.sciencefocus.com/science/can-you-unboil-an-egg