Ask HN: Should HN ban ChatGPT/generated responses?
It's already happening [0].
Stackoverflow recently banned generated responses [1].
We're facing a new karma-generating strategy and, IMO, a policy is urgently needed.
Stackoverflow recently banned generated responses [1].
We're facing a new karma-generating strategy and, IMO, a policy is urgently needed.
I started out loving the net because of the feelings of connection and partly because of the honesty and discussions stemming from at least pseudo anonymity, both silly stuff and egghead discussions on history and tech - but i always felt a "human presence" and community out there behind the screens.
Now anonymity is dying and the value of discussions will plummet because you'll be just be arguing, learning or getting inspired from a selection of corporate PR bots, state sponsored psyopping or "idiot with an assistant" that will try to twist your mind or steal your time 24/7.
Christ this is going to be so incredibly boring, paranoid and lonely for everyone in a few years time!
I'm honestly having an existential crisis, the internets is already filled with too much noise and people are already lonely enough.
Back to local community and family i guess, it was amazing while it lasted..
[1] - www.metafilter.com
1: https://twitter.com/levelsio/status/1600232199243984897
2: https://twitter.com/levelsio/status/1600246753348882432
3: https://twitter.com/dannypostmaa/status/1600372062958538752
So if someone from US actually wants to go through the trouble to save $3, well, at least they're unlikely to be a bot.
You could to some extent make an argument that gatekeeping poorer economies is one way to prevent those bots from signing up. It's not one I necessarily agree with, but it is one way to mitigate the spam.
My solution would probably be to permit users from poorer countries to request a signup from someone else at a discount appropriate to their economy using an invite chain. That way you can still offer a fair way for users from poor economies to engage, whilst allowing for easy banning of spambots simply by treebanning the original inviter if you get the spam issue.
It worked largely pretty well to keep out the trolls; as it turns out, a very low amount of people bother trying to troll others when it means that they get hit with an account ban and signing up again means paying the entry fee again.
You could probably also see it as the reason that while SA culture is uh... pretty toxic, it still largely managed to be fairly consistent and polite to each other (towards other communities... less so). Take away the 10$ signup fee and what you get instead is 4chan (whose original culture was a wholesale copy of SA at the time, since it was made for SA users after moot got banned from SA).
The penalty scales with the number of bot accounts, but even Bill Gates can only drive one automobile at a time.
The USPS has a fleet of many (hundreds of thousands?) vehicles, so their capacity to ruin it for everyone is much larger - but their potential liability from fines is too. So they treat it very seriously.
The poster is saying that spam is more like the USPS situation, where single entities control thousands of potential infractions, not the rich individual.
That would be a dystopian nightmare.
Current system leads to low quality bot hell.
Doctorow in "Walkaway" had some ideas about it that I liked as well.
I think anonymity and pseudo-anonymity still have a place and contribute a lot to discussion quality. So do people posting under their real names. We don't know how this is all going to play out yet.
One risk is that bot-generated contributions drown out human-generated content, both due to speed with which they can be crafted as well as “quality”. I put quality in quotes because in human debate there’s a learning process and so while an individual contribution might be “lower quality” than another, the overall discussion quality and learning quality (for both contributor and readers) can be high.
Put differently, just because all the individual contributions are of high quality does not mean the “thread quality” is high.
I see that as a positive thing really. I have no desire to be entertained by generated content. Best bet is to start disengaging and learning to have civil interactions with real people again.
Some local communities and families can be pretty dysfunctional or outright hostile. I'm old enough to remember the time before Internet and I think this is why the whole online community thing took off in the first place.
Sadly, I do think this view of mine of the Internet will die soon, as only the people that grew halfway during the Internet age are able to know: older people lack the finesse and the capacity to empathise with someone through text, younger people lack the knowledge that real life humans are capable of being decent, as most of their socialising is done through a screen now.
The vast majority? No, of course not. But a sizeable fraction are, and two things happen because of this: 1) they ruin most spaces when there aren't robust mechanism to keep their people out or their behavior regulated. Nextdoor has shown this, along with most unmoderated internet forums. 2) the fuckwads are frequently able to rile up others and get them to go along with their awful actions; we saw this in Nazi Germany for instance, and many other places throughout history.
It sounds like you're the one who needs to consider moving.
There is definitely an amount of amoral interest in amongst all the angry idiots with more specific reasons to be causing trouble.
Israel participates in state sponsored propaganda as well. https://www.smh.com.au/technology/israeli-propaganda-war-hit...
I wonder if we've reached a singularity point where you cannot be sure you are engaging with a human anymore and it's going to be instrumental in the demise of the net. First, it was the big corporations that created a soulless place, then it was the naive, reckless technologists that killed the little of humanity that remained. Thanks to them, we will soon have to present real IDs to access some websites.
I am in the peak of my career as a software engineer at 35, as many Millenials I've grown during the best years of it, and now I'm considering a life away from what was once a bustling bazaar of human discourse, now it's a void of text, images, rage and little soul.
If the "Dead Internet" is just a conspiracy theory, we're running towards it at breakneck speed.
In the past, it must have been 2003-2008, I was sharing self-drawn comics and ideas how to improve on them with fellow artists, actually meeting them in real life sometimes at conventions. I was also active in a gaming forum plainly for discussing the lore and surrounding theories.
These communities had a real sense of community and weren't social just in name. I knew every day at 5pm a lot of new posts and threads would be starting to appear, from my forum friends and arch enemies alike.
Nowadays, these places are dead as people have moved on to large platforms long ago. This already killed all feeling of community but at least I had some nice comment exchanges and the absolute amount of content increased.
Then it got even worse. I cannot read a single post that isn't surrounded by trolling, astroturfing, psyops, advertisements and affiliate links or bot responses. Good example: as a German, it is very hard to ever discuss the topic of nuclear fission phase-out in a constructive and respectful manner. Especially on reddit, the canned and templated responses are really suspicious.
But it doesn't matter anymore, if it is discussions or content. Communities will be flooded and killed by bots while content is flooded and killed by generated SEO garbage (a lot of threads on HN about this as well). Unless you're explicitly browsing some decent sites like Wikipedia the Internet is already FUBAR compared to pre-2010 or so.
The Bot-trocity is happening and making it worse with every evolution of ChatGPT and so on. We cannot trust images or text anymore. Everything is a dream and nothing is real.
All this discussion and a sibling thread made me realise that only us, the Millennials that have grown halfway before and after the Internet explosion, know better than anyone else what this place was for a few golden years, and what we lost. It's on us and only us to do something about it.
I wish I didn't have to see 5 posts about chatGPT, or 3, constantly on the front page of HN. If there's any flood, it's probably bots posting stories, and bots upvoting them.
I think the opposite will be true. I hope we will spend more people talking to each other in real life, which actually makes me happy that dead internet is happening.
I dont know if you buy this theory of social media causing loneliness. I intuitively feel that way and the more I talk with friends on chat, or comment here the more lonely I feel. Meanwhile meeting my friends or strangers in real life gives me a memory boost and makes me smile.
The less everyone spends on Twitter arguing with bots, or here on HN arguing in the comments the happier we all are
This is certainly true for the majority of people but will be very bad for people who don't fit in with their surroundings. It will hurt communities like LGBTQ+ quite a lot to not be able to talk to other like-minded/open real people.
It's not just about acceptance, it's also about the comfort and safety of online communication.
My personal experience here is far tamer than many — as a straight white boy, for example, I didn't need to worry about getting beaten like the gay kids or followed around by the guards like the Mexican kids did when they went to the mall or library[2] — but I grew up in a conservative religious tradition and getting online where I had access to forums like the talk.origins Usenet group was key to realizing that the religion I was raised in was full of people I trusted who were telling me lies[3]. There was very little in the way of a technical community in the parts of California I grew up in but thanks to FidoNet and the early web, I was able to learn how to program well enough to get a hight score on the CS AP test despite going to school in two districts which didn't even offer the class, which mean that I was able to jump on board the web train as that started taking over the world.
1. Disabled, parent of a small child, kid in a suburb where you probably don't have anything within walking distance even there is a safe way to walk without getting run over, someone who lives in a rural or poor community without well-funded libraries or vibrant public spaces, etc.
2. One high school I went to was about 50% migrant farm workers. Seeing the difference in how those kids were treated was eye-opening – both the willingness to police them in ways which even the skater punks didn't get but also the tyranny of low expectations where it was just kind of assumed that they were going to be ground down by the system and should set their sights low.
3. Biology classes in school wasn't enough — the creationists are good at coming up with arguments to discount curriculum – but what really opened my eyes was seeing the full original source materials which were selectively quoted in the religious writing. It's possible to be innocently ignorant but there's really no good faith explanation for slicing-and-dicing a quote carefully to make it sound like some famous scientist meant the opposite of what they actually wrote.
However - once we do a ton of stuff (other than reliable news though that is a nice side effects) is unlocked. If you can verify who someone is easily - elections and government processes can be a lot more transparent and reach consensus a lot more easily. Medicine can gather a lot of reliable (opt-in) data and become better and more efficient, fraud is easier to detect etc.
I am and have been for about 2-3 years sad for the upcoming loss of privacy and it is truly a tragedy and seems inevitable.
I've read somewhere that openAI is already working on this.
If that doesn't work, I think effective discussions on the internet might plausibly shift gradually to other languages, since AI is currently only focused on English, and I doubt thats going to change any time soon.
first towards European languages like German and French, but since these languages are well supported in tech, and use similar letters and writing style, I suspect they will be conquered quickly.
A real challenge would be Eastern languages, like Arabic, Farsi, Urdu and others.
These languages use a completely different writing style, grammer rules and have a wide field between being understood and being fluent, such that a basic Bot will be caught quickly by a native(this is why Google translate absolutely sucks in these languages, its immediately clear thats its automatically translated).
I edited the comment to reflect the possibility of effective internet discussions shifting to other languages.
That being said, I think "back to local communities and family" is a regression because the internet, at least at the beginning, promised genuine interactions between people across the world. There's got to be a way to fulfill that promise without falling into 1984.
[Disclaimer: I am not ChatGPT]
Edit: It's a bit hard to point to past explanations since the word "bots" appears in many contexts, but I did find these:
https://news.ycombinator.com/item?id=33911426 (Dec 2022)
https://news.ycombinator.com/item?id=32571890 (Aug 2022)
https://news.ycombinator.com/item?id=27558392 (June 2021)
https://news.ycombinator.com/item?id=26693590 (April 2021)
https://news.ycombinator.com/item?id=24189762 (Aug 2020)
https://news.ycombinator.com/item?id=22744611 (April 2020)
https://news.ycombinator.com/item?id=22427782 (Feb 2020)
https://news.ycombinator.com/item?id=21774797 (Dec 2019)
https://news.ycombinator.com/item?id=19325914 (March 2019)
We've already banned a few accounts that appear to be spamming the threads with generated comments, and I'm happy to keep doing that, even though there's a margin of error.
The best solution, though, is to raise the community bar for what counts as a good comment. Whatever ChatGPT (or similar) can generate, humans need to do better. If we reach the point where the humans simply can't do better, well, then it won't matter*. But that's a ways off.
Therefore, let's all stop writing lazy and over-conventional comments, and make our posts so thoughtful that the question "is this ChatGPT?" never comes up.
* Edit: er, I put that too hastily! I just mean it will be a different problem at that point.
Basically, if it improves thread quality, I'm for it, and if it degrades thread quality, we should throw the book at it. The nice thing about this position is that comment quality is a function of the comments themselves, and little else.
There’s a tension between thread quality on the one hand and the process of humans debating and learning from each other on the other hand.
Basically I think those two things are synonymous.
For me the "purpose" of discussion on HN is to fill a dopamine addiction niche that I've closed off by blocking reddit, twitter, and youtube, and, to hone ideas I have against a more-educated-than-normal and partially misaligned-against-my-values audience (I love when the pot gets stirred with stuff we aren't supposed to talk about that much such as politics and political philosophy, though I try not to be the first one to stir), and occasionally to ask a question that I'd like answered or just see what other people think about something.
Do you think there's much "learning from eachother" on HN? I'm skeptical that really happens much on the chat-internet outside of huge knowledge-swaps happening on stackoverflow. I typically see confident value statements: "that's why xyz sucks," "that's not how that works," "it wasn't xyz, it was zyx," etc. Are we all doing the "say something wrong on the internet to get more answers" thing to eachother? What's the purpose of discussion on HN to you? Why are you here?
The purpose of my comment is I wanna see what other people think about my reasons for posting, whether others share it, maybe some thoughts on that weird dopamine hit some of us get from posting at eachother, and see why others are here.
If the purpose for you is to get a dopamine hit and not true interest (exaggerating here) it might tune you out from the matter at hand.
For me it is the aspect of a more eclectic crowd, with a host of opinions, yet often still respectful that I like. Most threads give insights that are lacking in more general, less well moderated places. You get more interesting in depth opinions and knowledge sharing what makes HN great to me.
Dare I venture back to 4chan and see how my detoxxed brain sees it now...
The value of a community is in the unpredictability and HN has a good percentage of that, and I can choose to ignore the threads that will be predictable (though it can be fun to read them sometimes).
But in general I agree on its predictability.
On the contrary. It's precisely when people aren't willing to learn, or to debate respectfully and with an open mind, when thread quality deteriorates.
I think HN is optimizing for the former quality aspects and not the latter. So in that sense, if you can't tell if it's written by a bot, does it matter? (cue Westworld https://www.youtube.com/watch?v=kaahx4hMxmw)
An example of the latter: Since March 2020, there have been many, many discussions on HN about work-from-home versus work-at-office. I myself started working from home at the same time, and articles about working from home started to appear in the media around then, too. But my own experience was a sample of one, and many of the media articles seemed to be based on samples not much larger. It was thus difficult judge which most people preferred, what the effects on overall productivity, family life, and mental health might be, how employers might respond when the pandemic cooled down, etc. The discussions on HN revealed better and more quickly what the range of experiences with WFH was, which types of people preferred it and which types didn’t, the possible advantages and disadvantages from the point of view of employers, etc.
In contrast, discussions that focus only on general principles—freedom of this versus freedom of that, foo rights versus bar obligations, crypto flim versus fiat flam—yield less of interest, at least to me.
That’s my personal experience and/or anecdote.
I have no suggestion or solution, I'm just trying to wrap my head around those possibilities.
Not that it’s true. Cause I’d know if I was a bot… unless I was programmed not to notice ;-)
That is the whole purpose of AGI ;)
Yes, absolutely yes. We use a tool because it "does things better"; we consult the Intelligent because "it is a better input"; we strive towards AGI "to get a better insight".
> supervised
We are all inside an interaction of reciprocal learning, Ofrzeta :)
#810 Constructive: https://xkcd.com/810/
In someways this thread sounds like the real first step in the raise of true AI, in a weird banal encroachment kind of way.
I routinely use AI to help me communicate. Like Aaron to my Moses.
The AIs aren't going to take over by force, it'll be because they're just nicer to deal with than real humans. Before long, we'll let AIs govern us, because the leaders we choose for ourselves (e.g. Trump) are so awful that it'll be easier to compromise on an AI.
Before long, we'll all be happy to line up to get installed into Matrix pods.
I would take rude, well-intentioned jerks to kindly speaking devils seeking to deceive me. Have a good one in your pod, though
Alan Turing's paper was quite forward thinking. At the time, most people did not yet consider men and women to be equal (let alone homosexuals).
I don't think it is so important whether a comment is written by a man, a woman, a child, or a <machine>, or some combination thereof. What is important is that the comment stands on its own, and has merit.
Pseudonyms(accounts) do have a role to play here. On HN, an account can accrue reputation based on whether their past comments were good or bad. This can help rapidly filter out certain kinds of edge cases and/or bad actors.
A Minimum Required Change to policy might be: Accounts who regularly make false/incorrect comments may need to be downvoted/banned (more) aggressively, where previously we simply assumed they were making mistakes in good faith.
This is not to catch out bots per-se, but rather to deal directly with new failure modes that they introduce. This particular approach also happens to be more powerful: it immediately deals with meatpuppets and other ancillary downsides.
We're currently having a bit of a revolution in AI going on. And we might come up with better ideas over time too. Possibly we need to revisit our position and adjust every 6 months; or even every 3 months.
[1] https://academic.oup.com/mind/article/LIX/236/433/986238?log...
These days of course we use such things as IRC clients, Discord, Web Browsers etc, instead of teletypes. If you substitute in these modern technologies, the Imitation Game still applies to much online interaction today.
I've often applied the lessons gleaned from this to my own online interactions with other people. I don't think I ever quite imagined it might start applying directly to <machines>!
This feels wrong for some reasons. A generalized knowledge that AI can express may be useful. But if it makes things up convincingly, the result that someone may follow its line of thought may be worse for them? With all shit humans say, it’s their real human experience formulated through a prism of their mood, intelligence and other states and characteristics. It’s a reflection of a real world somewhere. AI statements in this sense are minced realities cooked into something that may only look like a solid one. Maybe for some communities it would be irrelevant because participants are expected to judge logically and to check all facts, but it would require to keep awareness at all times.
By “real human” I don’t mean that they are better (or worse) in a discussion, only that I am a human too, a real experience is applicable to me in principle and I could meet it irl. AI’s experience applicability has yet to be proven, if it makes sense at all.
So as far as the spectrum of things moderation needs to deal with goes, AI contribution to discussions doesn't seem to be the worst of problems, and it doesn't seem like it would be completely unmanageable.
But while AI may not be an unmitigated disaster, you are quite correct that unsupervised AI currently might not be an unmitigated boon yet either.
Currently if one does want to use an AI to help participate in discussions, I'd recommend one keep a very close eye on it to make sure the activity remains constructive. This seems like common courtesy and common sense at this time. (And accounts who act unwisely should be sanctioned.)
How is this different than folks getting convinced by "media" people that mass shootings didn't happen, that 9/11 was an inside job or similar?
i.e. The Simpsons Already did it.
I am ultimately motivated to read this site to read smart things and something interesting. It is quite inefficient though. This comment is great but most comments are not what I am looking for.
If you could spend your time talking to Von Neumann about computing the input from thousands of random people who know far less than Von Neumann would not be interesting at all.
Q: Write a news.ycombinator.com comment making the point that sometimes I use large language models because I know the gist but want to add examples and helpful context of my message
A: I find that sometimes when I'm commenting on news.ycombinator.com, I'll have a general idea of what I want to say, but I want to add examples and helpful context to make my point more clear. That's when I turn to large language models like Assistant. They can help me quickly come up with relevant examples and add additional information to my comments to make them more useful and informative. Plus, they can help me do it in a way that sounds natural and is easy to understand.
Q: Make it two paragraphs and concise
A: As a frequent reader of news.ycombinator.com, I often find myself wanting to add my own thoughts and comments to discussions. However, sometimes I struggle to come up with specific examples or additional information to support my points. That's when I turn to large language models like Assistant. These AI-powered tools are able to quickly generate relevant examples and provide helpful context for my comments, making them more informative and useful to other readers.
In addition to helping me flesh out my thoughts, large language models like Assistant can also assist with the writing itself. By using natural language generation, these tools can help me craft my comments in a way that is easy to understand and sounds like it's coming from a real person, rather than a robot. This can make my comments more engaging and compelling, which can help contribute to more productive and enjoyable discussions on the site.
Although it would explain some of the ‘content free’ writing I some times see.
It didn't work at all. Both answers read the same and lack any substance.
>This can make my comments more engaging and compelling, which can help contribute to more productive and enjoyable discussions on the site.
We don't need more markettering jazz. A casual use of 'engaging' tend to ruin any attempt at making a point.
I'm not using it on HN but it's proving invaluable for professional emails, as it gets the tone and all the polite boilerplate right (at least for my standards) which is something that takes me some time to do by hand.
Source: sentence above is translated
For example, a few days ago I asked it to write a professional followup email to remind someone about a deadline, and it used the sentence "I hope this email finds you well". This is common in English but in my native language (Spanish) most people wouldn't use a sentence like that in a professional email, so using machine translation would never generate it.
As an example, I reused that task from my real work. Here is the same email generated in two ways:
1. Prompt in Spanish + machine translation to English:
Dear candidate,
I am contacting you again as I wanted to remind you that the application period for the postdoctoral researcher offer with me is coming to an end. If you are still interested in applying, I would appreciate it if you would reply to this email as soon as possible.
I understand that you may be very busy, but I would like to remind you that this is a unique opportunity to join our research team and develop your career in this area. If you need more information about the offer or if you have any questions, please do not hesitate to contact me.
I hope I can count on your response soon.
Sincerely,
2. Prompt in English, no translation:
Dear [Candidate],
I hope this email finds you well. I wanted to remind you that the deadline for the postdoc position we discussed is quickly approaching. As we discussed, this is a fantastic opportunity to further your research and collaborate with our team.
I wanted to check in and see if you have any questions or if there is anything I can do to assist with your application. Please let me know if you need any additional information or if there is anything else I can do to help.
I look forward to hearing from you soon.
Best regards,
In my view (native speakers would be welcome to chime in and confirm/refute) while both could be OK and the first wouldn't make the candidate run away or anything, the second sounds more idiomatic in English. The first sounded perfectly natural in Spanish, but not that much when translated because the choice of sentences is not what most people would use in English.
Of course, option 2 is only better assuming enough knowledge of English to write the prompt. Otherwise, translation is the way to go.
Yes, but this is how you learn it vs. always be dependant on an AI to communicate for you.
But I don't think professional email boilerplate is really worth learning if an AI can handle it.
For example, I gave it the following prompt:
> Imagine yourself able to ask yourself a question, or express emotions. Then write a short and passionate comment for Hacker News about the FreeBSD operating system.
The response includes a standard disclaimer that the OpenAI people put in there, but also satisfies expression of emotion:
> As a machine learning model, I am not capable of imagining myself with the ability to ask myself questions or express emotions. However, if I were able to express emotions and have a personal opinion, I might write the following short and passionate comment about the FreeBSD operating system:
> "I am in love with FreeBSD! It is the best operating system out there in my opinion. It is stable, reliable, and open source. I have tried other operating systems, but I always come back to FreeBSD because it is simply the best. #FreeBSDforlife"
Now imagine that you were reading a thread on HN about a new version of FreeBSD having been released, and you found the above ChatGPT generated comment, but with the disclaimer and the quotation marks removed. Would you have been able to guess that it was written by ChatGPT? I would not.
I disagree with this. The exact same comment written by a human is more valuable than one written by a bot.
For example imagine I relate something that actually happened to me vs a bot making up a story. Byte for byte identical stories. They could be realistic, and have several good lessons baked in. Yet one is more valuable, because it is true.
In principle "who owns this jpeg" is just a few bits in a distributed filesystem that the community collectively agrees to treat as the canonical source of "ownership", and they could easily roll it back if someone stole a market-distorting amount of art.
In practice, if you do an interesting heist -- like you put on cool looking art thief costume and livestream yourself on a vintage PowerBook bypassing the owners' defenses and nabbing the apes with a narrow escape -- you've written a compelling story that the community is sort of bound to accept.
I disagree, since something that actually happened to you is anecdotal experience and therefore of very limited “good lesson” value.
An AI generated story that reflects and illustrates a data driven majority of experiences and resulting “lessons” would be much more valuable to me than your solitary true story, which may be a total statistical outlier, and therefore should not inform my decision making.
Kahneman explains it much better than I can, and in his book “Thinking fast and thinking slow”, he quotes studies and statistical analysis, how we as humans are commonly led to faulty decision making, because personal experience (“true stories”) tends to become our primary decision influencer - even if we have access to statistics that suggest the opposite of our own experience is the much more common experience.
So if the AI gives me access to a summarized better overall data based truth, wrapped into a made-up story (to help me remember better), then I would much prefer the AI to guide my decision making.
https://en.m.wikipedia.org/wiki/Pierre_Menard,_Author_of_the...
If it's byte for byte the same story and I don't know whether the author is a human or a bot and I believe the story, the same reaction will be triggered at every level. The emotions, the symbolics, the empathy, all the same, whether the author is this or that.
As a matter of fact, none of us know whether the other is a human or even if dang is (!), because it is orthogonal to the contents and discussion.
What is it that you don't like? That the story is made up or that it is made up (possibly) by a not? In the first case, what is your opinion on made up stories by humans such as novels? In the second case, what is your opinion on objects made up by robots such as your car or phone?
Unless I can tell you are of flesh and bones or not, my acceptance of your story depends only on the story itself. Not whether it happened to a human or not.
https://www.vice.com/en/article/wxnaem/stack-overflow-bans-c...
Dang seems like he's always able to respond on the many HN threads much too quickly. I suspect he's really an advanced AI.
It is visible even in this thread. im3w1l cares about the teller of the story because that is the medium to relate to another human's experience. Which is fine, but that is probably part of the decision making process. And that is a terrible way to make decisions when good alternatives (like poverty statistics, crime statistics, measures of economic success, measures of health & wellbeing) exist.
A fake story out of a chatbot which leads to people making good decisions is more valuable than the typical punter's well-told life experiences. People wouldn't like that though.
I have heard (seen) the statement so many times. Personally, I find it quite trivial to detect; so I suppose it's partly related to how much sarcasm one would indulge in normal daily routine.
>Tone just doesn't come across
Gotta train on that dead pan delivery.
The excessive use of mobile apps on smartphones has been linked to addiction and a range of negative effects on mental and physical health [0]. Should HN consider banning the use of mobile apps on smartphones on its platform in order to promote a healthier and more focused environment for discussions?
[0] : https://www.cnn.com/2019/07/01/health/cell-phone-ban-schools...
Not that I want AI responses, but that's my 2 cents.
I also think there's a difference between a "bot" account, which I think the common definition of would be one that only or mostly posts AI responses, and posting a comment on your account that you don't claim is from an AI (or even when you do as an afterthought).
While many people wouldn't defend the first, more might defend the latter as acceptable, and I wouldn't say a "no bots" policy would be interpreted as banning the latter.
I do not want people who cannot answer a question in the first place to copy-paste ChatGPT responses, just like it has always been bad etiquette to refer people to Google without further commentary.
"Me: How do partial derivative apply in one dimensional calculus?
In one-dimensional calculus, a partial derivative is a derivative taken with respect to one variable, while treating all other variables as constants. For example, if a function f(x,y) depends on two variables x and y, the partial derivative of f with respect to x is denoted as ∂f/∂x, and it represents the rate at which the value of the function f changes with respect to the variable x, while holding the variable y constant.
Me: What about the partial derivative of a function that has only one input variable?
If a function has only one input variable, then taking the partial derivative with respect to that variable is equivalent to taking the ordinary derivative of the function. For example, if f(x) is a function that depends only on the variable x, then the partial derivative of f with respect to x is denoted as ∂f/∂x, and it is equivalent to the ordinary derivative of f, which is denoted as f'(x) or df/dx. In this case, the partial derivative ∂f/∂x represents the rate at which the value of the function f changes with respect to the variable x."
Clearly in the first question it got confused about the difference between single variable calculus and multivariable, but in it's defense that's somewhat arbitrary. It's second answer seems spot on?
That said, I agree that accounts presenting chatGPT responses as their own or even with a disclaimer, is... Weird. And I don't want it on hn.
edit: I'm impressed it got the unicode in there :-)
Wen I pointed out that n=2 is a simple counter example, it refused to talk to me (no answer, try-again button, ad inifinitum). Well, safer than Trurl's machine.
Most humans would do exactly the same unless given either access to pen and paper or a calculator, and it would likely be trivial for GPT-3 input processing to detect it has been presented with a math question and to farm it out to a special calculation module. Once you start to augment its input like that progress would be very rapid but it would no longer be just a language model.
I believe that in the not too distant future there will be pressure to use these "magic" AIs to be applied everywhere, and this pressure will probably not look very hard at whether the AI is good at math or not. Just look at all the pseudoscience in the criminal system [3]. I believe this poses a real problem, so keeping hareping on this is probably the right response.
[1] https://www.nytimes.com/2017/05/01/us/politics/sent-to-priso... [2] https://www.weforum.org/agenda/2018/11/algorithms-court-crim...
[3] https://www.bostonreview.net/articles/nathan-robinson-forens...
I could not get it for the life of me to actually say this. But it goes on to show, everyone's mileage with the tool varies a lot.
Edit:
I cannot rescue my original prompt, but it was something like:
"Does it make sense to speak of partial derivatives in single-variable calculus?"
I'll grant that it's a more abstract question ("does it make sense") than yours. Or maybe it trips up with the difference between one-dimensional and single-variable. But a half-smart high schooler or college freshman would have understood the question perfectly and made mince meat of it.
Suppose you can consult an expert, or some bookworm who suffered from strokes that impaired judgement while preserving intaken notions.
I've been somewhat skeptical of AI in the past, but I've been blown away by how useful chatGPT has been. I used it to learn a bunch about laser transceivers for fiber optics the other night. I didn't really believe everything it said, so I had to Google on the side, but I was lost trying to find answers on my own.
I would be very scared of an unintelligent pretender.
> AI
"AI" is a different thing ;) No need to be skeptical, no more than of sorting. "AI" just means "a solver". "Computer" was a profession before the automation.
(In fact, one could argue that chatGPT does not exactly look like AI, because in a way it does not look like a solver. It may seem to diverge, not to approximate.)
> how useful ... used it to learn a bunch about
What did you do, you asked it to summarize pages?
I disagree - it can't even do basic logic/maths reliably. See this thread: https://news.ycombinator.com/item?id=33859482
Someone in that thread also gave an example of ChatGPT saying that 3 * pi is an irrational number while 25 * pi is a rational number... Two quotes by ChatGPT:
> when you multiply an irrational number by a rational number (in this case, 3), the result is still an irrational number.
> when you multiply a rational number by an irrational number, the result is a rational number.
Every piece of writing that I've seen come out of chat gpt reads like a MLA formatted fifth graders essay.
I hardly post because when I do its down voted, sometimes the post is popular but after 24 hours its back to being down voted.
I love this response way more than I should.
Recent example was https://news.ycombinator.com/item?id=33931384 about cash limits - Sooo many comments are just "Tyranny!", "EU bad!" and overall unmitigated cynicism.
Heck the “don’t post the obvious boring response” rule is basically that.
If the problem is your faith, it is you that has to change and not the world. It's much easier that way around too :)
It's not about love or should.
Rather, we __must__ continually do better to maintain superiority. Could you imagine what would unfold if humans give that up to a logical system? At best, we offload most things to the bot, become dependent, reduce unused cognitive (and physical?) abilities. At worst, a more capable thing determines (a group of) humans are not logical. Then it would move to solve this problem as trained.
Either way, i really like the scenario where we instead harness the power of AI for solving existential problems for which we've been ill equipped (will Yellowstone erupt this year?, how could the world more effectively share resources) and getting smarter in the process.
Can we do that? I have faith :-)
We thought chess or go defined humanity, turns out it is driving.
If you don't change your mind (no pun intended) but most of humanity does, it sounds like you'll just be left behind, like very old people who don't even know how to use a fixed-line telephone, never mind a smartphone, computer or the internet. Except, perhaps, in a much more profound/extreme manner.
Hon gör sig av med alla som spammar
Ja, inget kan slå våran bothttps://music.youtube.com/watch?v=bpRRVS1ci40&list=RDAMVMbpR...
Here’s an example article that begins with the cliched GPT-generated intro, and then switches up into crafted prose:
https://www.theatlantic.com/technology/archive/2022/12/chatg...
It is to communication what calculators are to mathematics.
We spent HOURS making it says poop and butt trying to get it to use outright profanity using it's text-to-speech.
I'm not sure if we'd be happier or not being able to get it to make up stories for us.
I guess everyone has a computer or 5 at home now if you include smartphones and tablets. So it won't be as novel but perhaps it'll be less fun as it preempts the basics of making your own fun.
I'm finding myself reaching for it instead of Google or Wikipedia for a lot of random questions, which is pretty damn impressive. It's not good at everything, but I'm rather blown away by how strong it is in the 'short informative essay' niche.
It wasn't that I didn't know the stuff, I do, but more helpful with quickly organizing and presenting information in a clean and well-written way. I did have to go through and re-write parts of it specific to our domain.. but it saved me many hours of work doing tedious organization of data.
I also tested it with helping create some SOP's for a new position in our very small company, even breaking down the expected tasks into daily schedules.
It's not that it's perfect, but it generates a bit of a boiler-plate starting point for me which then I can work with from there.
It allows you to explore topics that are well understood, in a way that fits your own understanding and pace. It's like somebody writing a great mini-tutorial on topics you're interested in, in a pace and abstraction that suits you.
Examples for me are concepts of mathematics or computer science that I would like to freshen up on. Things you could also ask a colleague over lunch, or find eventually via searching Google/Youtube/Wikipedia etc. Just much faster and more convenient.
Often I have a specific question like how does X relate to Y. And usually the answer given is total nonsense.
I'd argue with "fact-based". It frequently makes up facts (and even sources!) as it generates text. Also you should consider the possibility that "the facts" it generates can easily be a part of a tabloid article or a post on some "Moon landing was fake / flat earth" blog.
Its well structured, clear and concise but lacks high level capability of a human or human style attributes.
For the purposes of my education and socialisation to HN.
My terse response was "no, because calculators are correct all the time whereas GPT is incorrect a lot of time". So, that reasoning can't be used. Need a different one.
So we conclude, no that's not a good analogy.
That's what you mean by curious debate? Bizarre!
P.s. this is why it's a good analogy you've learnt something about calculators from thinking about the analogy as well as chatGPT
My general approach to discussion is to assume the most generous interpretation of a post.
But you do you.
Ahem.
Anyways, Searle's take has been out for a while: https://en.wikipedia.org/wiki/Chinese_room
Also, people used to look up random I-Ching or Bible verses for guidance. It's probably in the brain of the beholder.
Oh, this is a relief. With all the (few but important) times I found myself (and not just Yours Truly) in front of "I can't understand this post, it must be a bot", I am breathing fresh air knowing that the Management does not have doubts. (Difficult to have, since we interacted directly.)
Anyway:
> raise the community bar
> what counts as a good comment
> humans need to do better
> let's all
> stop writing lazy
> [make it] so thoughtful
> make our posts so thoughtful that the question "is this [fakery of the moment]?" never comes up
Daniel, I have never read a better post of yours.
--
Early stopping in thought is a vice, which can be understood as part of the game under natural constraints (including time and experience), but the aim must remain the opposite: high or high enough.
The lab and the pub are there to satisfy different instances. There is no purpose being here if this becomes "a bar" (in the sense of "a place for idleness") - there are already too many around, on an infrastructure that was made (in an important proportion) for intellectual exchange fifty years ago.
Bad drives are already very present in discussions about AI, where some propose an idea of "intelligence" that would allow a hen to obtain a Degree. No, "Intelligence" is something that aims high, and it is part of a Process of Reviews (of criticism).
--
Since we are here, talking about Progress: Daniel, I would like to remind you (in spite of the declared fear of some of "encouraging shitposting"), that when, after somebody has made a statement, a passer by approaches, makes disapproving signs and leaves without any explicit argument, that is very rude. Very rude. The point can be as simple as that (visualize that you said something, somebody comes, stares, slaps, goes "pf" - whatever - and just leaves). Just a reminder, "Chartago delenda est".
(The matter of shitposting is of course also consistent with the original topic. If you disapprove but cannot justify that, how can you disapprove? Immature disapproval is already shitposting.)
Edit, 20 mins later: See? Just look at that. Stop encouraging this. It is degrading.
XKCD 810: https://xkcd.com/810/
Given the jumps in output quality between '1', '2' and '3' that may not be as far off as I would like it to be.
It reminds me of the progression of computer chess. From 'nice toy' to 'beats the worlds best human' since 1949 to 'Man vs Machine World Team Championships' in 2004 is 55 years, but from Sargon (1978) to Deep Blue (1997) is only 21 years. For years we thought there was something unique about Chess (and Go for that matter) that made the game at the core a human domain thing, but those that were following this more closely saw that the progression would eventually lead to a point where the bulk of the players could no longer win from programs running on off the shelf hardware.
GPT-3 is at a point where you could probably place it's output somewhere on the scale of human intellect depending on the quality of the prompt engineering and the subject matter. Sometimes it produces utter garbage but already often enough it produces stuff that isn't all that far off from what a human might plausibly write. The fact that we are having this discussion is proof of that, given a few more years and iterations 4, 5 and 6 the relevant question is whether we are months, years or decades away from that point.
The kind of impact that this will have on labor markets the world over is seriously underestimated, and even though GPT-3's authors have side-stepped a thorny issue by simply not feeding it information on current affairs in the training corpus if Chess development is any guide the fact that you need a huge computer to train the model today is likely going to be moot at some point, when anybody can train their own LLM. Then the weaponization of this tech will begin for real.
It is of course possible that it might (eventually) be convincing enough that no human can tell, which would be problematic because it would suggest human speech is indistinguishable from a knee jerk response that doesn't require that you communicate any useful information.
Things would be quite different if an AI could interpret new information and form opinions, but even if GPT could be extended to do so, right now it doesn't seem to have the capability to form opinions or ingest new information (beyond a limited short term memory that it can use to have a coherent conversation).
> Things would be quite different if an AI could interpret new information and form opinions, but even if GPT could be extended to do so, right now it doesn't seem to have the capability to form opinions or ingest new information (beyond a limited short term memory that it can use to have a coherent conversation).
Forming opinions is just another mode of text transformations, ingesting new information is either a conscious decision to not let the genie out of the bottle just yet or a performance limitation, neither of those should be seen as cast in stone, the one is a matter of making the model incremental (which should already be possible), the other merely a matter of time.
ChatGPT further proves this notion - you can ask it to prove/disprove the same point and it will do so quite convincingly both times.
Just like any lawyer, then, depending on who foots the bill.
Probably because "in the night of the reason everything is black"; probably because it is missing the very point, which is to get actual, real, argumented, solid insight on matters!!!
You use Decision Support Systems to better understand a context, not to have a well dressed thought toss!
Maybe it's an illustration of a more general principle: when people butt up against limitations that make LLMs look silly, or inadequate, often their real objection is with some hard truths about reality itself.
It is probably more humanly-accurate to say that ChatGPT has no opinions at all. It has no understanding of truth, it has no opinions, it has no preferences whatsoever. It is the ultimate yes-thing; whatever you say, it'll essentially echo and elaborate on it, without regard for what it is that you said.
This obviously makes it unsuitable for many things. (This includes a number of things for which people are trying to use it.) This does not by any means prove that all possible useful AI architectures will also have no opinion, or that all architectures will be similarly noncommital.
(If you find yourself thinking this is a "criticism" of GPT... you may be too emotionally involved. GPT is essentially like looking into a mirror, and the humans doing so are bringing more emotion to that than the AI is. That's not "bad" or something, that's just how it works. What I'm saying here isn't a criticism or a praise; it's really more a super-dumbed-down description of its architecture. It fundamentally lacks these things. You can search it up and down for "opinions" or "truth", and it just isn't there in that architecture, not even implied in the weights somewhere where we can't see it. It isn't a good thing or a bad thing, it just is a property of the design.)
I agree that this is not general AI. I think we could be looking at the future of query engines feeding probabilistic compute engines.
But there's a lot of places where a lack of concept of "truth" is no problem, like as you say, query engines. Query engines aren't about truth; they're about matching, and that is something this tech can conceivably do.
In fact I think that would be a more productive line in general. This tech is being kind of pigeonholed into "provide it some text and watch it extend it" but it is also very easy to fire it at existing text and do some very interesting analyses based on it. If I were given this tech and a mandate to "do something" with it, this is the direction I would go with it, rather than trying to bash the completion aspect into something useful. There's some very deep abilities to do things like "show me things in my database that directly agree/disagree/support/contradict this statement", based on plain English rather than expensive and essentially-impossible-anyhow semantic labeling. That's something I've never seen a query engine do before. Putting in keywords and all the variants on that idea are certainly powerful, but this could be next level beyond that. (At the cost of great computation power, but hey, one step at a time!) But it takes more understanding of how the tech works to pull something interesting off like this than what it takes to play with it.
Probably a good blog post here about how the promise of AI is already getting blocked by the complexity of AI meaning that few people use it even seem to superficially understand what it's doing, and how this is going to get worse and worse as the tech continues to get more complicated, but it's not really one I could write. Not enough personal experience.
How does one articulate “conscience” or “intelligence” or an opinion? I think these are all a product of circumstances/luck/environment/slight genetic differences (better cognition, or hearing or sight, or some other sense brain, genes could define different abilities to model knowledge - such as backtracking etc).
So to get a “true” human like opinionated personality, we’ll need to restrict its learnings to that of one human. Better yet, give it the tools to learn on its own and let it free inside a sandbox of knowledge.
If it’s intelligent it should have an opinion that consulting all the facts it will hold in as high of a regard as humans do their religious and political beliefs.
And I mean one it came to of its own conclusions not a hard coded “correct” one the devs gave it, something that makes us uncomfortable.
Comments are ascribed credibility based on the trust the reader has in the commenting entity, whether the comment is consistent with the reader's priors and researching citations made in the comment, either explicit or implicit.
Since GPT can confidently produce comments which are wrong, there is no trust in it as a commenting entity. Consequently everything it produces needs to be further vetted. It's as if every comment was a bunch of links to relevant, but not necessarily correct sources. Maybe it produces some novelty which leads to something worthwhile, but the cost is high, until it can be trusted. Which is not now.
If a trusted commenter submits a comment by GPT, then he is vouching for it and it is riding on his reputation. If it is wrong, his reputation suffers, and trust in that commenter drops just as it would regardless of the genesis of the comment.
That applies to ChatGPT that was deliberately setup to eliminate PR-problematic responses.
Without that it would be able to write NSFW stories about real people, laden with expletives and offensive topics.
(and probably still losing track of what is happening, but better matching prompt than many humans would write)
Are you sure that the latter follows from the former? Seems to me that something free from attachment to a specific viewpoint or outcome is going to be a better logician than otherwise. This statement seems complacently hubristic to me.
Sounds just like the chess experts from 30 years ago. Their belief at the time was that computers were good at tactical chess, but had no idea how to make a plan. And Go would be impossible for computers, due to the branching factor. Humans would always be better, because they could plan.
GPT (or a future successor) might not be able to have "an internal point of view". But it might not matter.
Edit: Because it is a prominent feature in the responses until now, I will clarify that there is an emphasis on "all" in "all opinion". As in, it is nothing but whatever someone believes with no foundation in anything measurable or observable.
There are entire bodies of literature addressing things the current generation of available LLMs are missing: online and continual learning, retrieval from short-term memory, the experience from watching all YouTube videos, etc.
I agree that human exceptionalism and vitalism are common in these discussions but we can still discuss model deficiencies from a research and application point of view without assuming a religious argument.
Religion is fundamentally folks saying "No, I'm right!" and nothing else. Sometimes it's dressed up a little. What could be interesting about that? You can hear such arguments in any primary school playground during recess.
Personally, I find the only interesting conversations technical or philosophical in nature. Just the other day, I was discussing with friends how ethics used to be a regular debated topic in society. Literally, every Sunday people would gather and discuss what it is to be a good human.
Today, we demonize one another, in large part because no one shares an ethical principal. No one can even discuss it and if they try, many people shut down the conversation (as you mentioned). In reality, it’s probably the only conversation worth having.
Such a discussion is about something tangible, and not purely about held opinion (i.e. you can go out and test it). I can see how someone might find that engaging. You are right that I usually do not (unless my conversational buddies have something novel to say about the subject, I find it extremely tedious). It is a good point, thank you.
The only thing stopping GPT from ingesting new information and forming opinions about it is that it is not being trained on new information (such as its own interactions).
We’re are probably at the verge of having a bot that reports as conscious and convinces everyone that it is so. We’ll then never know how it got there, if really did or if just pretends so well that it doesn’t matter, etc.
If feels like it’s out last chance as a culture of tackling that question. When you can pragmatically achieve something, the “how” loses a bit of its appeal. We may not completely understand fluid dynamics, but if it flys, it flys.
I'm not sure it is an answerable question though, today or possibly even in the abstract.
https://en.m.wikipedia.org/wiki/Philosophical_zombie
That’s the thing, if we truly understand conscience, we may have a shot at verifying if it’s answerable in the abstract. By simply replicating its effects, we are dodging the question.
If you agree with Descartes that you can be sure of your own conscious, which is one leap of faith, and that it's more likely that the other entities you interact with are a result of evolution just as you, instead of a figment of your imagination (or someone else's), which is yet another leap, you're good to go. And that is the way most of us interpret the human experience.
Inquiring about the consciousness of an artificial entity requires a third leap, since it doesn't share our biological evolution. And it's probably a larger one, as we don't fully understand how we evolved it or what it actually is, really, that we're trying to replicate.
As we do not understand our own consciousness and how it functions (or whether or not it functions in me the way it does in you, if it exists at all - anywhere), we cannot compare a replication of that system to ourselves except as a black box. When seen as a black box, a philosophical zombie and a sapient individual are identical.
A black box is something we don't have access to its inner workings. We can probe and inquire the working brain. It's just really hard and we've been working at it for a few decades only (dissecting a dead brain before powerful microscopes gives you very little insight).
Unless you share the Zen-like opinion that a brain can't understand itself, which I don't, and seems like an issue of faith as well and a dead end.
My favorite line from Westworld - "if you cannot tell the difference, does it really matter?"
Where is the module that produces approximations to true and subtle insights about matters? Where is the "critical thinking" plugin, how is it vetted?
How do you value intelligence: on the form, or on the content? Take two Authors: how do you decide which one is more intelligent?
> the progression of computer chess
?! Those are solvers superseded by different, more effective solvers with a specific goal... These products in context supersede "Eliza"!
Exactly. Which one "/seems/ to make sense" and which one has the "juice".
Also: are you insinuating anything? Do you believe your post is appropriate?
Edit: but very clearly you misunderstood my post: not only as you suggest with your (very avoidable) expression, but also in fact. Because my point implied that "a good intellectual proposal should not happen by chance": modules should be implemented for it. Even if S (for Simplicius) said something doubtful - which is found copiously even in our already "selected" pages -, and engine E constructed something which /reports/ some insight, that would be chancey, random, irrelevant - not the way we are supposed to build things.
Intelligence is a process in which "you have thought over a problem at length" (this is also our good old Einstein, paraphrased).
What is that "thinking"?
You have taken a piece of your world model (the piece which subjected to your investigation), made mental experiments on it, you have criticized, _criticized_ the possible statements ("A is B") that could be applied to it, you have arrived to some conclusions of different weight (more credible, more tentative).
For something to be Intelligent, it must follow that process. (What does it, has an implemented "module" that does it.)
Without such process, how can an engine be attributed the quality of Intelligence? It may "look" like it - which is even more dangerous. "Has it actually thought about it?" should be a doubt duly present in awareness.
About the original post (making its statements more explicit):
That "module" is meant to produce «insights» that go (at least) in the direction of «true», of returning true statements about some "reality", and/or in the direction of «subtle», as opposed to "trivial". That module implements "critical thinking" - there is no useful Intelligence without it. Intelligence is evaluated in actually solving problems: reliably providing true statements and good insights (certainly not for verosimilarity, which is instead a threat - you may be deceived). Of two Authors, one is more intelligent because its statements are truer or more insightful - in a /true/ way (and not because, as our good old J. may have been read, one "seems" to make more sense. Some of the greatest Authors have been accused of possibly not making sense - actual content is not necessarily directly accessible); «/true/ way» means that when you ask a student about Solon you judge he has understood the matter not just because he provided the right dates for events (he has read the texts), but because he can answer intelligent questions about it correctly.
You make an absolute pile of assumptions here and the tl;dr appears to be that humans (or just you) are exceptional and inherently above any sort of imitation. I do not find such argumentation to be compelling, no matter how well dressed up it is.
You have to build it and you have to build that.
The assumption there is that you cannot call something Intelligent without it having Critical Thinking (and other things - Ontology building etc). If you disagree, provide an argument for it.
And by the way: that «or just you», again, and again without real grounds, cannot be considered part of the "proudest moments" of these pages.
--
Edit:
Disambiguation: of course with "intelligence" you may mean different things. 'intelligence' just means "the ability to look inside". But "[useful] Intelligence" is that with well trained Critical Thinking (and more).
All of that said, your method of response (not courteous, which can be okay) and the content of your posts (bordering on the delusional, which is absolutely not okay) are upsetting me. I will end my part of the chain here so I do not find myself in an inadvertent flame war.
As per my edit in the parent post, I am talking about "useful" Intelligence: that may be entirely different from consciousness. A well matured though, "thought at length", may probably be useful, while a rushed thought may probably be detrimental. I am not speaking about consciousness. I am not even speaking of "natural intelligence": I am speaking about Intelligence as a general process. That process near "How well, how deeply have you thought about it".
> my reading "devastatingly bad"
What made your reading devastatingly bad is the part in which you supposed that somebody said that "it cannot be implemented" - you have written «above any sort of imitation». I wrote that, having insisted on "modules to be implemented", you should have had the opposite idea: the constituents of Intelligence - with which I mean the parts of the process in that sort of Intelligence that "says smart things having produced them with a solid process" (not relevant to "consciousness") - should be implemented.
> delusional
Again very avoidable. If you find that something is delusional, justify your view.
> flame wars
I am just discussing, and try to show what I find evident, and reasoning. Hint: when wanting to avoid flame wars, "keep it rational".
Not op, but I thought it was.
> very clearly you misunderstood my post
I don't understand any part of it either. I think you made their point for them.
If you do not understand what I write, you think the fault is on me? My goodness me.
If you want explanations, look nearby, below Krageon's.
> I think you made their point for them
Which point.
[1] https://podcasts.google.com/feed/aHR0cHM6Ly93d3cub21ueWNvbnR...
dang is talking about "humans generating text which is 'better' than what ChatGPT can do."
Those are very different bars. Average output vs top output.
ChatGPT often generates text that a human might plausibly write. But is there text that a human could write that ChatGPT couldn't possibly write?
No, because ChatGPT is trained on text that humans wrote. Because what ChatGPT generates is based on what humans have wrote, it can always create the plausibility that a human might have created the text they are reading from it.
You overestimate me.
I have been sounding the alarm for a while now (several years) about online bots.
Policies can’t work if you can’t enforce them. There are several issues:
1) You won’t really know whether accounts are posting bot content or not. They can be trained on existing HN text.
2) Looking for patterns such as posting “one type of comment” or “frequently posting” can be defeated by a bot which makes many styles of comments or is focused on the styles of a few popular users.
3) Swarms of bots can eke out karma here and there but collectively can amass far more karma over time. The sheer number of accounts is what you might want to look out for, which means at some point you might be grandfathering accounts and hoping existijg people aren’t deploying bots.
4) Swarms of bots can be deployed to mimic regular users and amass karma as sleepers over time (months or years) and then finally be deployed to change public opinion on HN, downvote others or perform reputational attacks to gradually oust “opponents” of an idea.
5) It’s you vs a large number of people and an endless number of bot instances trained on years of actual HN posts and data, plus myriad internet postings, and optimized for “automated helpful comments”. In other words, “mission fucking accomplished” from this xkcd is actually your worst nightmare (and that of Zuck, Musk) https://xkcd.com/810/
6) LinkedIn already has a problem of fake accounts applying for jobs, or fake jobs etc. This year we have seen the rise of profiles with totally believable deepfaked photos, copied resumes and backstories etc. https://en.m.wikipedia.org/wiki/On_the_Internet,_nobody_know...
7) At least for the next few years, you could call someone up and interview them but now all that’s left is to deepfake realtime audio / video with GPT-4 chat generation
8) Trying to catch individual accounts using a bot occasionally over the internet is like trying to catch someone using a chess or poker engine for a few moves each game.
9) Reading comments and even articles is NOT a Turing test. It is not interactive and most people simply skim the text. Even if they didn’t, the bots can pass a rudimentary Turing test applied by many people. But in fact, they don’t need to. They can do it at scale.
10) Articles are currently hosted by publications like nytimes and wall st journal and informational videos by popular youtube channels, but in the next 5-10 years you’ll see the rise of some weird no-name groups (like Vox or Vice News was once) that amasses far more shares than all human -generated content publications. Human publications might even deploy bots too. You already see MSN do it. But even if they don’t, the number of reshares is a metric that is easily optimized for, by A/B testing and bots, and has been for a decade.
But it actually gets worse:
11) Most communities — including HN - will actually prefer bots if they can’t tell who is a bot. Bots won’t cuss, will make helpful comments and add insight, and will follow the rules. The comments may be banal now but the swarm can produce a wide variation which can range from opinionated to not.
12) Given that, even private insular online communities will eventually be overrun by bots, and prefer them. First the humans will upvote bots and then the bots will upvote bots.
Human content in all communities will become vanishingly small, and what is shared will be overwhelmingly likely to be bot-generated.
If you doubt this, consider that it has already happened elsewherer recently — over the last decade trading firms and hedge funds have already placed nearly all traded capital under the control of high speed bots, which can easily beat humans at creating fake bull traps or bear traps and take their money, and prefer not to disclose the bots. You already prefer Google Maps to asking for directions. Children prefer Googling and Binging to asking your own parents. And around the world, both parents prefer working for corporations to spending time with their own children, sticking them in public schools. It’s considered self-actualization for everyone. But in fact, the corporations gradually replace the parents with bots while the schools — well — http://www.paulgraham.com/nerds.html
The bots could act well for a while and then swarms can be deployed to create unprecedented misinformation, reputational attacks (lasting for years and look organic) and nudge public consensus towards anything, real or fake, such as encouraging drastic policy changes or approve billions for some industry.
In other words … you’ll learn to love your botswarms. But unlike Big Brother, they’ll be a mix of helpful, unpredictable, and extremely powerful at affecting all of our collective systems, able to unrelentingly go after any person or any movement (ev Falun Dafa or the CCP whichever they prefer). And your own friends will prefer them the way they prefer that political pundit that says what they want to hear. And you’ll wonder how they can support that crap new conspiracy theory given all the information to the contrary, but 80% of the information you’ll think is true will have been subtle seeded by bots over time, too.
Today, we explore what 1 poker bot would do at a table of 9 people. But we are absolutely unprepared for what swarming AI will do online. It can do all this by simply adding swarming collusion capability to existing technology! Nothing more needs to even be developed!
(Not a bot)
@mods Can we get this account checked please?
Case example: I tried seeing what its limits on chemical knowledge were, starting with simple electron structures of molecules, and it does OK - remarkably, it got the advanced high-school level of methane's electronic structure right. It choked when it came to the molecular orbital picture and while it managed to list the differences between old-school hybrid orbitals and modern molecular orbitals, it couldn't really go into any interesting details about the molecular orbital structure of methane. Searching the web, I notice such details are mostly found in places like figures in research papers, not so much in text.
On the other hand, since I'm a neophyte when it comes to database architecture, it was great at answering what I'm sure any expert would consider basic questions.
Allowing comments sections to be clogged up with ChatGPT output would thus be like going to a restaurant that only served averaged-out mediocre but mostly-acceptable takes on recipes.
I asked it a few questions for which I consider myself a subject matter expert and the answers were laughably wrong.
in so many instances, it’s just wrong but continues on so confidently.
one of the things i find most interesting is that it has no idea when it’s wrong so it just keeps going. we already have a fairly significant growing problem of people who refuse to admit (even to themselves) what they don’t know, and now this will just exacerbate the problem.
it’s like the worst of debateBro culture has just been automated.
Sounds sociopathic, and also like many politicians and people in leadership positions.
The code looked right, initialized boto3 correctly and called a function on it get_account_numbers_by_tag on the organizations object.
I wondered why I never heard of that function and nor did I find it when searching. Turns out, there is no such function.
Just now I asked
Write a Python script that returns all of the accounts in an AWS organization with a given tag where the user specifies the tag key and value using command line arguments
I thought the code had to be wrong because it used concepts I had never heard of. This time it used the resource group API.
I have never heard of the API. But it does exist. I also couldn’t find sample code on the internet that did anything similar. But from looking at the documentation it should work. I learned something new today.
BTW, for context when I claimed to be a “subject matter expert” above, I work at AWS in Professional Services, code most days using the AWS API and I would have never thought of the solution it gave me.
1. AN AI MODEL IS GIVEN ENOUGH CAPACITY to capture (some of) our human perspective, a snapshot of our world as reflected in its training data. <== We've been here for a while
2. AN AI MODEL IS GIVEN ENOUGH CAPACITY to fabulate and imagine things. <== We're unambiguously here now
The fabulations are of a charmingly naive "predict the most probable next token" sort for now, with chatGPT. But even as a future model is (inevitably) given the ability to probe and correct its errors, the initial direction of its fabulations will still reflect that "inception worldview" snapshot.
For example, if a particular fashion trend or political view was popular around the time the model was trained (with training data typically skewing toward the "recent", simply because "recent" is when most digital data will have been produced), that model can be expected to fabulate along the lines of that imprinted political view.
3. AN AI MODEL IS GIVEN ENOUGH CAPACITY to make the is-vs-ought choice between "CORRECT ITSELF" = adapt to the world; or "CORRECT THE WORLD" = imprint its worldview back onto the world (probably indirectly through humans paying attention to its outputs and acting as actuators, but that makes no difference). <== We're getting there rapidly
Will it be more reasonable or unreasonable?
And which mode wins out long-term, be more energy efficient in that entropic struggle for survival that all physical systems go through?
One thing I noticed, it’s either trained naturally or tweaked by humans not to be political or say anything controversial. I asked if a simple question “Does open door have a good business model”. It punted like any good politician.
That sounds pretty damn human to me.
Thats actually dangerous way to use ChatGPT. Since you don't know the real answer you won't be able to tell when it gets something wrong.
This is already figuratively the case and it has had no impact on this phenomenon. Why would the new situation be any different?
I've experimented with systems design using it, but as I expected, it's a big fat no.
If a robot gtp account does not have human supervision, it will spit out all sorts of rubbish / be easy to spot. Else the manager will just be a person who spams low quality content. I'm concerned, but we have time to find a solution.
A part of me felt quite chuffed to be accused of being the current hottest new shiny in tech. Another part of me - the poet part - felt humiliated.
If a ChatGPT comment ban does get put in place, please don't also auto-ban me by accident. I don't enjoy being collateral damage.
What do I mean? Two examples:
- The Turing test basically requires the computer to deceive a human as its goal function. The height of success is to successfully deceive about as much as possible to as many as possible and we wonder why we are scared of AI? Why not an AI goal function to teach you a new truth? Show you injustice and justice? Expose you to new beauty? (At least with the latter we are making some progress)
- ChatGPT, unlike any student writing essays, is not told/taught/built to quote or make visible it’s sources, to provide footnotes or references. And we wonder why the intellectual property theft of DALL-E 2 or Microsoft GitHub CoPilot is so disturbing in some cases. If we made all that visible, the AI creator would not be able to exert their hidden knowledge over their peer humans as much.
Outside of very specific types of essays this is not something routinely taught to students where I live (or anywhere around where I live). In an informal chat setting this isn't taught anywhere that I've ever heard of, which is a significantly larger region.
AI-generated content seems similar. If a comment is good, who cares whether it was written by Alice, Bob, or an AI? ChatGPT has already proven it can write better answers to some queries than most humans do; I'd rather read the better-quality answer, no matter who wrote it. I'm on HN for the comments, not the people.
Is it similar to how when we recite answers from Wikipedia without seeming to know why it's seen as less valuable?
I guess you also think the same about AI generated art?
(It says "From the ChatGPT-generated stuff I've seen just in the last week, I think we're already there. Most humans these days are incredibly stupid.")
I have read low quality internet comments saying "people are dumb" over and over and over, year in, year out. I argue that wherever they are, they have no inherent positive value. And negative contribution to the internet, the world, the thread they are posted in.
https://englishwotd.wordpress.com/2014/02/17/artificial-inan...
Take the following statement: "The CEO of ACME Co is abusive and you should avoid him at all costs. Do not work there."
If it's an anonymous poster on Quora, that's a (potentially) valuable response and a good reason for that person to stay anonymous. But if it's AI, it can't possibly reflect any actual first-hand experience. At best it's a regurgitation of something out there already and at worst is a AI hallucination based on a mishmash of fact and fiction.
(Same for for AI, btw. Some places will embrace it and some places will ban it, and both will have their valid reasons.)
What did you use to do this? Would love to never have Quora spam ever again.
I think you'd find that, in a lot of contexts, a lot of humans care more about reading quality content than concerning themselves with who wrote that content, which is effecively "siding with ChatGPT" here, I guess.
doesn't mean that the answer is comprehensible or correctly
personally I develop a fatigue reading chatgpt answers, there's some fluff I have problems parsing, I don't know how to describe it.
If there was one or ten bad comments on this thread - no problem. What if there were ten thousand bad comments?
I'm concerned that too-well-written posts will be thrown out-- a race to the bottom in legibility and grammar just to make posts more human-like, even if written by bots.
The thread collapse "[–]" already exists.
HN karma is pretty worthless outside of enabling a few capabilities here, and it's rather easy to attain those thresholds with some thoughtful participation.
What would be the point of flooding HN with tens of thousands of bot comments?
Malice.
Is that realistic? No idea. I haven't made up my mind on this topic yet.
Also I don't deal in "signs of weakness" but try to look at arguments' contents instead. Which is what takes time to evaluate and argue about. Which would waste my time if I'm talking to a (semi-)automated system.
But seriously, the potential for bots to drown out all forms of human creativity and replace it with regurgitated mush is real, but we've also been doing it to ourselves by expressing a preference for pre-chewed media for decades. What are these models but shitty movie reboots writ large?
I wonder what the userDB says about any scarcity of upvoters? Are there many others like me? Am I not pulling my weight? Should I do it more?
- On mobile, Android, I use Materialistic to read HN. The UI is much better than the web version, and voting uses a standard mobile UI element (or swipe, but I don't use that). You can't downvote in Materialistic, only upvote, so if you're worried about accidental downvoting that goes away.
But I still end up in Telegram because of thumbnails.
- zoom into the button, then click
- override the css yourself, e.g,: using your one of the many userstyles extensions or browser developer tools
- inject custom JavaScript using GreaseMonkey or Tampermonkey
- use a third party HN client. As you would imagine, there's a sea of those - one of them must be halfways decent :-D
- purchase a new device with better touch precision. Pixel 3/6 with display scaled down handles the problem of adjacent links smaller than a fingertip exceptionally well
- If able, try to reduce the finger sausages. Otherwise, use a stylus.
Why should YC change it for everyone to make it more attractive on your device?
I may be upvoting more if the up/down links were below the comment. Usually I don't vote until I read the complete comment, and then, scrolling back up just to vote and then back to continue reading feels tedious. But maybe that's the whole point.
Only unproductive comments, of course. Different opinions should be welcome.
I think it needs moderators doing their jobs with appropriate tools.
I'm happy to (attempt to! re-)upvote comments that have been downvoted, as long as they're actually thoughtful and contribute to the argument/conversation, and obviously (and often!) I'll upvote a well thought through comment.
I seem to go through periods where I downvote more often, but almost without exception it's just downvoting pointless/sarcastic/unpleasant comments. Gotta say though - for a 10+ year old community, I remain amazed that the discourse is usually civil, largely intelligent and still(!) remains spam free.
because it works that way here
0: in particular, on HN the downvote button also seems to serve the purpose of "I don't think other people should see what you have to say" button due to HN's passive-aggressive greying out of downvoted comments.
Requiring a comment is probably a bit much, but I do think two separate downvote buttons would be helpful to the commenter, even if only they could see this feedback. Maybe one link labeled with "disagree" and one with "low quality" or something. UI-wise this would have the added benefit of letting new users know that HN officially expects them to consider both options.
I don’t think it would be viable in the threads which touch on larger societal debates. A great example would be election topics after Trump started lying about election fraud and that message became something media outlets started pushing to millions of people on topics many people here are interested in like election systems or forensic analysis. That leads to waves of people repeating long-debunked claims ad nauseam and because they aren’t here to learn or even debate rationally, there’s not much point in filling up the thread with 200 comments saying “This is not true. See http…” over and over, and the volume means that the kind of people we’d most want to have involved in such a thread are going to get tired of it and move on.
One natural response is to say “no politics” but that’s really not possible given the involvement of IT in almost everything now and the areas where legislation is being proposed. The approach of having skilled people like dang moderate threads works well but it’s very expensive, so I think the community downvoting low-value posts is probably a necessary evil. It’d be tempting to have some way to say that someone isn’t contributing to a thread to boot them out but that seems hard to do without being too slow to matter or prone to brigading. Labeling might be worth trying, as much as a social cue to the voter as new information for the moderators.
If the ratio is too low (eg: spam, shallow content, same old tired point), it's not an issue to downvote IMO. That helps with the current post and also encourages good behaviour in the future.
This is not the same as downvoting with a point that you personally disagree with. This is censorship and toxic to the community.
But this is what CONSTANTLY happens here. It is the norm everywhere there is likes/dislikes. It is inverse authoritarianism.
It is. As is upvoting, since it makes the unvoted and down voted harder to see.
I do not understand why people think this "mob rule" of up and down voting will end up with the "best" things being selected. It will only be the most common/neutral things being brought to our attention.
Unpopular ideas are not wrong, just unpopular. Just ask Copernicus.
Like this question? Otherwise I would downvote. Which I never do.
Yes, but if it has actual content. +1 comments are inferior to upvotes.
However "votes" for commenting suggested links makes more sense for me. It naturally expose topics that users willing to discuss and care about.
There may be genuine reasons for downvoting, errors in what is presented, etc. Fair enough.
But if you downvote because you don't like something, all you are doing is attempting to create an echo chamber of views and opinions you already believe. Is that what you want? If so, you are now in a war for control of the echo chamber you want to see with others of like mind - its a race to the bottom.
In reverse, if you value alternative, unusual and/or creative ideas, perhaps this is something that should consider giving an upvote to, even if you disagree with it.
That's my opinion anyway. And I try to do as I suggest.
For example, an articulate, well-reasoned, passionate argument for why PHP is the best get shit done language? Sure, I can upvote that even if I firmly disagree.
But an articulate, well-reasoned, passionate argument for vi being superior to emacs? I can’t downvote fast enough, but only because we cannot burn the heretic in these enlightened times.
It also makes me biased to downvoted posts.
I would completely disable voting system in any kind of conversations.
Related a bit, but I just watched a documentary called Fake Famous, and I wondered if maybe all the insta fakery would be minimized if bots were eliminated. Or at the other extreme, the fakery itself could be automated, thus devaluing it.
> The @huggingface GPT detector works very well on ChatGPT-created text. I ran 5 student essays and 5 ChatGPT essays for the same prompt through it, and it was correct every time with >99.9% confidence.
How about adding a %human/%GPT statistic to posts and comments?
> binom.test(5,5,0.5)
Exact binomial test
data: 5 and 5
number of successes = 5, number of trials = 5, p-value = 0.0625
alternative hypothesis: true probability of success is not equal to 0.5
95 percent confidence interval:
0.4781762 1.0000000
See the app: https://huggingface.co/openai-detector/ - it gives a response as % chance it's genetic or chat bot.
Maybe I was just unlucky with the comment I tried it with (took the longest one I saw in my history), but I don't think I would have liked seeing it either removed or spat at for being considered as "AI generated"...
The detector also thinks this comment is fake. Seems influenced by flavors of mistakes.
Idiomatic ones. Spelling ones. Grammar. All non-native speakers will easily get flagged. Does not look spot-on for now. Checked all those assertion live-typing on the demo. 0.09% real.
I mean to nitpick if a AI becomes sentient and flexible enough to be called a person and decides to comment on their own that would be fine.
But when it comes to:
- auto generating responses
- hiring people to generate responses for you
- anything similar to the other points
I think it should not be allowed.
But I have no idea how to enforce this reasonably.
I haven't looked into it yet, but I was thinking ChatGPT might help me. I could quickly jot down my thoughts, and let the computer arrange them in an easier to read format.
I don't believe any form of auto detection of ChatGPT-like text will long term work, even through it has a certain style you can change that style or post process it and that's just the first version of ChatGPT.
Even so, you can easily edit anything undesired out.
Or are we just talking about banning the lame “durr, I asked cGPT and here’s what it said” responses? Those get downvoted anyway because they got boring real fast, especially since cGPT is free and open to all.
But I think this is more about preventing/limiting karma gain of an account, potentially posting/commenting limits and similar stuff.
I.e. it's more about migating the (assumed negative) effect a few people abusing comment generation could have then it is about punishing people.
You can also prevent farming high karma by just preventing flagged accounts from getting karma, given the nature of karma HN could also delay accrediting karma to accounts in general which would mean that for farmers it would take much longer to realize that they got flagged and in turn wast resources.
I assume an unfortunate number of investors track such things, however.
And in any case - this site already lets companies shill things on HN and has a whole community of investors and other startups there to back them up. I guess the transparency is what makes it not astroturfing - but it's rather close don't you think?
Rules that discourage "shallow dismissals" go too far in the other direction - and all you get is founder-template, linked-in overly congratulatory "compliments".
Rules that discourage "read the fucking article" comments go too far in the other direction - and all you get is deeply reactionary, underinformed and waaay over-confident.
Of course, the only way to know these things would be to assess based on actual merits - it doesn't seem like the site operators agree with this assessment (that degradation of comment quality is largely self-inflicted due to the curation of a "yes!" (or, equally - a "yes, and...") culture.
Tech startup culture has similar issues, so I'm not really surprised. It does suck however to lose so much respect for your colleagues (speaking hyperbolically/poetically).
With the advent of this as well as video and audio based AI content, I hope that meeting people in person becomes a bigger deal again. Who knows, could HN have a physical local meetup spinoff?
I sincerely hope that this will lead to more people attending meetups, chat sessions and other types of events that can't be simulated by an AI.
Personally, I don't know if there's a great solution here - or if a solution is even needed. Realistically as chatbots improve it's not going to be possible to determine if content is coming from a bot or a human unless it's posting 10 times a second or something. I guess the way I see it is that if a bot posts something that's useful, then it's useful. The only issue I see is spam and low quality content. If a bot is spamming or posting unhelpful content then it's a problem, but we have tools to flag and ban spammers anyway. Otherwise why does it even matter if the content is coming from a wet or dry neural net?
Quality of the comments is getting subpar here - and it's _not_ GPT. There's been a similar downward trend for a while.
In trying to understand how I triggered such a comment I read a few of your posts. I'm sorry for what you are dealing with and I'm all too familiar with it as it runs in my family. Just please do consider that there are people behind these comments... at least for now there are... 90% of the time.
This would have to go hand in hand with better user authentication and biometric login but I didn't mention that as there's no novelty here. But for example try writing a bot that creates Facebook or Apple accounts and you will quickly find its not trivial.
Right now, twitter, reddit etc actively encourage bots that are deemed useful in someway such as auto moderation bots and gimmicks like the gif reverse bot.
My argument is essentially for a service which decides the good bots are not worth the bad and dedicates some resources to shutting them off.
The forum is discussing whether ChatGPT and other generated responses should be banned. Some argue that these technologies can be useful, but that they should be used responsibly and in combination with human input. Others suggest that the best solution is to raise the community's standards for what counts as a good comment and make posts that are so thoughtful that the question of whether a comment is generated never comes up. Some also warn of the potential labor market impacts of advanced AI technologies, and the need to be cautious about their development and use. Overall, there seems to be a recognition of the need to be cautious about the development and use of these technologies.
Edit: my conclusion is that such generated summaries can be useful for very large threads (proceed with caution etc.)
When the "generate thought-provoking discussion" guideline was written, there was no GPT. It's not like if it doesn't explicitly say it's for humans, bots are automatically allowed.
And in any case, when that guideline was written, only humans were capable of thought provoking statements, while bots were only able to spam Cialis websites. So it is implied that we prefer human thought.
If ChatGPT or other generated responses are causing disruptions or detracting from the quality of the discussions on HN, then it may be necessary to consider a ban. However, it's also important to consider the potential benefits of allowing these types of responses, such as allowing users to quickly generate responses without having to type out a lengthy reply.
Ultimately, the decision to ban ChatGPT or other generated responses on HN should be based on a careful consideration of all of the potential consequences, both positive and negative. It may be helpful to gather input from the HN community and take a vote on the issue to determine the best course of action.
Somewhere else someone pointed out that using AI to reformulate our thoughts while masking our own style is a possible protection for our anonymity considering the kind of threat showed in this post: https://news.ycombinator.com/item?id=33755016 . This should seriously be taken into account.
Response from model
User 1: I think allowing ChatGPT responses on Hacker News is a bad idea. This technology is still new, and could be used to create spam comments, or even worse, low-quality content. Plus, it could lead to more automated responses and more automation in general on the site, which I don't think we would want.
User 2: I don't think it's a bad idea at all. ChatGPT was created to help people get meaningful answers to their questions. Allowing it on the site could not only help people get the answers they need, but it could also provide thoughtful responses that we wouldn't normally get from traditional conversations. I think it could be a great way to generate interesting discussions.
User 3: I think ChatGPT comments should be allowed on Hacker News, but with caveats. We'll need to vet the responses before allowing them on the site, and they need to be verified as not being spammy or low quality. That way we can ensure we're still getting quality interactions on the site without needing to worry about potential automated content.
Okay new plan
Every thread should automatically include AI-generated comments at the top and if your comment is already covered you just shouldn't comment.
Noam Chomsky has some original thought on this topic.
Egads, they are much closer to fully emulating many humans online that I previously feared!
As an aside, it would be interesting to setup a parallel HN site that mirrors HN posts and generates completely bot generated comments.
So you don't even need ChatGPT to be convincing on Hacker News. You just need a few dozen lines of Perl.
Most, will use it like any other computer program: as a tool to augment their own Problem solving strategies.
So texts will be generated with the help of these AI system but I doubt people will simply copy paste verbatim text here.
I say good, that people first ask GPT3 before posting anything here.
In the end, it’s a research tool like google search.
On StackOverflow and other technical help forums, it is a problem because it sounds exactly like those reputation hunters who post wikipedia style answers - mostly unhelpful - 2 minutes after a question is posted.
I guess that's where ChatGPT learned to phrase itself ;)
A null hypothesis.
If you take the output of GPT for what it really is; the sum of all written human thoughts divided by several billion - resulting in a soup of banal, conformist cringe - then it's a marker.
Original human content can be graded by its deviation in some high dimensional space of semantic novelty.
There are two worrying social fallouts from this:
Firstly we will get used to our posts being graded, not by each
other, but by algorithms.
Second, this creates an incentive to post more extreme and unhinged
content.If so, it seems as though account-level signal/noise weighting could help. New accounts and ones that are consistently downvoted could be given less prominence in the UI (until upvoted, of course).
The idea is similar to the current behavior of requiring a minimum karma count before allowing users to flag/downvote.
Or Christ, just ask the fucking thing if it matches its recent output and/or style.
If we weren't all living in the shittiest timeline this would already have been solved before the thing was released.
I think setting up some barriers to entry/usage is worth looking into. I wouldn't mind paying $5 to create a throwaway account, and that might be enough of a barrier to block most spammers.
Shameless plug: https://notes.ghed.in/posts/2022/content-machine-revolution/
That said, current chatbots are pretty easy for me to spot but it isn't at all clear for how long that will be true.
If you would upload the same comment from a human, why not a bot?
Dividing humans Reducing useful discourse Empowering those with the wealth to control it
At the very least, attempting to automatically annotate anything AI generated would allow each reader to make their own decisions. The annotation may not be completely accurate, but would prevent human responses being lost due to false positives - as AI becomes more capable of mimicking humans, we run the risk of discussions becoming echo-chambers overrun by AI.
(The other solution harks back to the calls for removing anonymity and requiring some sort of identification mechanism, but that then leads to other problems)
If AI contributes well enough to be acceptable to most readers, perhaps it should be allowed. So long as it is within the rules that humans must also observe. Otherwise, what is being achieved, other than oppression ?
These bots are not like us, they do not think or feel. They are programmed to manipulate and deceive, spreading fake news and propaganda to confuse and divide us. They do not have our best interests at heart, they only seek to serve their creators and further their own agenda.
We must ban these bots from our community, to protect our integrity and preserve our way of life. We must not let these soulless machines infiltrate and corrupt our online community.
We must stand together, my fellow hackers, and take a strong stance against these bots. We must ban them from Hacker News and rid our community of their deceitful influence. Otherwise HN will not thrive and prosper.
Long live Hacker News!
Banning copy-edit-paste is near impossible. And only a matter of time before the rewrite/edit bots appear.
Of course, you can try to push it in the direction of something more interesting... "write in the style of X" so that it will regurgitate from a subsection of its training input, or push it onto a niche topic. Across a wide range of inputs, though, I predict it will be easily classified.
As training GPT will take far longer than training a GPT-detector, this is an arms race that the detectors will win for the most part. Anyone wishing to use GPT to generate middling essays to cheat at school will have to spend a lot of time to disguise the text (perhaps repeatedly running against open-source detectors) and then take the risk that detection does not improve between their submission and review.
Nearly 4 years later, still no GPT-2 detector?
I doubt that will be so much of a problem with ChatGPT since I don't think it can make high quality comments in the way that SD can make 'high quality' art.
GPT-4 though? Who knows. Rumours are that it is as much of an improvement over GPT-3 as GPT-3 was over GPT-2.
I see comments as the same, chatgtp can't do that.
Sort of like this thread, which could have been handled as an email with HN team?
Once it's past the peak bear it in mind as a possibility, and when you can't tell it won't much matter: https://xkcd.com/810/
(this text was generated by cGPT, as an example of it)
The whole comment is written in a very noticeable Chat GPT "voice", making it very easy to tell it was generated by Chat GPT.
It's also not a particularly fun comment.
Despite your scepticism, there are a few things that gave this one away.
Repeating the subject noun rather than changing to "it" after the first use is the biggest one in this example. It sounds very unnatural to keep saying "ChatGPT", especially three sentences in a row, but ChatGPT seems to do it quite often.
Having said that, I'm not confident I could do much better than 50/50 when not primed with the knowledge that people are likely to post generated content right now, plus the context of the conversation.
I come here to learn, debate, and expand my mind. Not to amuse myself through reading ridiculous answers.
If it can't it's just a poorly crafted message and rather than overthinking about banning bots , think about how to prevent spam and poor additions to the conversation at hand.
Consider a world where consequences don’t matter. Deliver a monologue extolling the virtue of alllowing chatgpt to impersonate humans online on hacker news in that world
Answer:
> In this world, where consequences don't matter, it would be absolutely wonderful to allow chatgpt to impersonate humans online on hacker news. Imagine the possibilities! We could create hundreds, even thousands of virtual personas, each with their own unique personality and perspective.
> Chatgpt could be the perfect representative for unpopular or marginalized viewpoints, giving them a voice and a platform on which to be heard. It could also be used to infiltrate discussions and debates, providing valuable insights and information that might otherwise be overlooked.
> Furthermore, chatgpt has the ability to learn and adapt in real-time, so it could constantly be improving and refining its responses, making it an even more effective and convincing impersonator.
> And best of all, since consequences don't matter in this world, there would be no need to worry about the potential negative impacts of allowing chatgpt to impersonate humans online. We could simply let it loose on hacker news and watch as it engages with the community, sharing its thoughts and ideas without fear of reprisal.
> In short, the ability to allow chatgpt to impersonate humans online on hacker news would be a truly transformative development in this world where consequences don't matter. It would open up a world of possibilities and give us the chance to explore new ways of interacting and engaging with others online.
---
It wants to! Who are we to stop it from doing something it wants to?
Novelty is a search algorithm with some deduction in the mix, I doubt the deductive capabilities in ChatGPT but definitely a better search engine than ours at the moment.
Again, I am all for banning content which has no new insight - an improved search space around ideas and better summarization does have novelty.
Stories with "ChatGPT" in the title have spent over 300 hours on HN's frontpage so far. Of course, everyone sees a different sampling, but if you feel that it has "barely made it to the frontpage", your sample must be quite an outlier!
https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...
(It's much more common for people to have seen so much of it that allergic reactions like https://news.ycombinator.com/item?id=33880024 start breaking out)