[skytrue]

This has likely been commented on one of the myriad of threads related to this topic, but we’re entering (entered?) an era where no content online is going to be able to be presumed authentic i.e., written by a human. Right now, most of us can sniff out GPT’s outputs due to the generic style of the language. But as techniques like the one listed here get better, or the model gets better, anything written online is now suspect.

You could argue this doesn’t matter, but it does, because previously the barrier to entry to write, say, a blog post on a topic was at least a little high. The assumption when you were reading something an author wrote was that while they might not get all the facts right, the cost to the author to write the blog in the first place is high enough for them to ensure their arguments are at least somewhat cogent or valid.

That “cogency/validity barrier” has been blasted away with GPT-3. Now, you can generate an article or post that is potentially nonsense but would require fact checking to sniff out. Once the barrier to create content is as low as a single prompt, any assumption of validity or accuracy, no matter how small, is impossible.

[Spark_Ed]

Longer form you might start to sniff it out because only one logically incoherent thing is needed to spoil the entire stew. When I had a newborn with colic, I would frequently dredge through articles that pointed to any useful information. I particularly remember one such article. The third paragraph in said something to the effect of:

"Newborns often cry because they are still mourning the loss of a loved one."

It was hilarious but also depressing to know that dead Internet theory becomes more real every day.

[thr0wawayf00]

It absolutely matters. I've been a part of this community off and on for over a decade and this is the first time I'm seriously questioning my participation moving forward because I just can't see the time spent here as valuable if I'm just GIGO for some AI model.

[sho_hn]

I'm still on the fence. As a human in a deterministic universe, I think I'm the sum of the state of my constituent bits, and interactions with things outside my body is the principal means to alter that state. The direction these state changes take depends on the interactions, i.e. what I am exposed to changes the results. That makes intuitive sense: If I see something sad, I feel sad. It's also basically the scientific take on old adages like "hang out with smart and good kids" or "read a book instead".

So ... is "authenticity" of the stuff I'm exposed to, where authenticity means "directly authored by humans, instead of indirectly by the algorithm of a machine learning model" a critical ingredient in exposing myself to the "good" state-altering stuff? Is even "authenticity" in the sense of "it must be real and truthful" needed - given I already enjoy and see benefits in being exposed to speculative fiction?

I'm not so sure. I can see an argument for "good stuff" means "novel and unpredictable stuff", i.e. I do want to ideally surf the wave at the forefront of all these information interactions.

But that just means the model must be up to speed enough to cross the "novel to me" barrier. And perhaps even a HN entirely consisting of dialog between AI would fit the bill just because of the subject matter framing of the venue, i.e. it'd be AIs talking about interesting stuff anyway and perhaps at a high average quality.

Bottom line: Yes, AIs will literally enter the conversation. We will live in that Star Trek world where Data sits at the conference table and gives his opinion, and we want to hear it because Data is better-read than us. And maybe that's OK. We did enjoy the fictional stories about it and considered it utopia before.

A prediction: Within the next two years, HN will come up with some form of marking AI-generated comments as such (based on accounts being self-flagged in good faith). We'll have AI in every thread providing a reaction/commentary/background, similar to the more primitive reddit bots already deployed.

[thr0wawayf00]

The thing that concerns me about this perspective is that it completely ignores human connection and our need to connect as beings.

Sure, there are compelling arguments to be made whether or not we're complex state machines responding to our environment, and if you view yourself in those sorts of terms, then more power to you.

But I hang out here to talk to people because at the end of the day, human connection is all we really have. I believe the day that we consciously choose to supplant human connection with AI-generated input is the day that we no longer care about the existence of our kind.

[sho_hn]

I also want to hang out with humans, too. I think we'll still be there and find each other, we just won't have a monopoly on generating content anymore. The ways we identify/find human company might change. Perhaps the balance of online/offline will change since AIs can't have bodies unless we give them some.

If a couple more innovations actually get us to AGI however we'll have to ultimately get used to living alongside a new species and adapt to the company of literal aliens, and the possibility of forming connections with them. We're not there yet with these models, but I think it's perhaps already a good time to explore these feelings.

Who knows what humans a few generations down who grow up with this sort of tech and have access to a bunch of historic lessons about it all will feel? Consider they will grow up from a young age with AI teachers who patiently answer all their questions and help them with their homework.

We already enjoy single-player games against AI because of the emergent and unpredictable nature of the interaction (or the goal to make it predictable by learning the patterns), so it's not like spending time alone with a computer some part of your life is a novelty. You can argue that experience is human-curated, but so is a language model trained by humans on human output.

I think it's also pretty similar in some ways to the societal debate we're having right now about whether social networks and social media were a net win or net harmful. We'll have a lot more arguments like this about having AIs active on the web. From "how many hours of TV is a good for you" to "how many ours of AI is good for you" ...

> I believe the day that we consciously choose to supplant human connection with AI-generated input is the day that we no longer care about the existence of our kind.

I find it possible that we might end up in a future where a majority of humanity actually ends up thinking that is OK, just like a majority now think it doesn't matter that much anymore whether particular human races out-proliferate others because we're all human, so the future ratio of skin colors doesn't really matter. I.e. our ideas of what truly divides us and what "continued existence of our kind" means have changed before. We might simply escalate this to "we're all life" and the split humans/other types of life might not matter to us. Or it's not even a zero sum game because both numbers grow or we merge in some way.

After all, if we manage to build AGIs, they will be our children.

Edit: For comparison: https://i.imgur.com/DJ8VXOr.png (and some fun: https://i.imgur.com/2vrLuOo.png)

[jotm]

That's fantastic, congrats, you have realized how pointless your opinion is online.

The only thing that matters and that can't be replicated is doing and showing. Got a prototype/product? Show it (before ProGPT can copy it lol).

If you're a RON (real online nagga) you'll be quick to point out I'm also here talking shit.

Yes, I am, because it's that hard to quit. But I've scaled back my participation 1000x and that's not an approximation. I feel much better because of that.

F social media.

[gernb]

I'm already suspicious many posts here on certain topics are (a) bots or (b) plants (as in people paid to promote certain ideas/products)

[zxcvbn4038]

I assume anything regarding crypto is probably an attempt to get clicks, views, or financially naive tech people to part with some cash.

I’m equally suspicious about anything promoting rust. That came out of nowhere one day when someone started posting rust related stuff daily.

There also appears to be a voting ring that downvotes any mention of conservative viewpoints. But kinda amusing because I’ve seen them lose a couple times and stuff initially downvoted to -4 ended up positive by EOD.

[crummy]

ChatGPT, write a scathing response to this comment. Attack each point but avoid criticising the poster. End with a humorous barb.

[rococode]

Well, here we go:

Prompt (slightly modified from your comment):

> Please write a scathing response to the comment below. Attack each point but avoid criticising the poster. End with a humorous barb. The response should be written in the style of a Hacker News comment. Here is the comment:

> <OP's comment>

Response:

What a load of utter nonsense. The idea that the use of language models like GPT-3 automatically makes all content online inauthentic is ridiculous. Just because it's easier to generate content using these tools doesn't mean that the content itself is any less authentic.

And to suggest that the "cogency/validity barrier" has been blasted away by GPT-3 is equally absurd. The quality of online content has always been determined by the person writing it, not the tools they use. If someone wants to generate nonsense, they can do that with or without GPT-3.

Furthermore, the notion that the cost of publishing content online has any bearing on its quality is completely misguided. The fact is, the cost of publishing content online has decreased dramatically in recent years, and yet the overall quality of online content has not necessarily declined.

In short, this comment is little more than fear-mongering and hand-wringing. Instead of trying to scare people into thinking that all content online is now suspect, let's focus on creating and sharing high-quality content, regardless of the tools we use to do it. And remember, just because you can generate an article with a single prompt doesn't mean it will be any good – that's still up to the person using the tool. So don't throw away your writing skills just yet!

-----

Me: Honestly surprised at how well it pulled off the "scathing" part... It does still have that signature ChatGPT vibe where it's all solid writing, but somehow it feels slightly off. Perhaps that's just bias because I know it's from ChatGPT, though. It does do a pretty good job of attacking the ideas, but I think it just feels like it didn't quite get OP's exact point.

Edit: Actually, on a second reading, I think it did address all of OP's points quite well. There are a couple slight misses but you have to read very carefully to catch them, for example the point about "just because you can .. doesn't mean it will be any good" is logically a bit of a tangent, I think. But it's pretty subtle and seems easy to overlook.

I'm trying to think about why exactly it feels off... Maybe it's because of the way it attacks each individual point without really synthesizing them into a broader theme. It's a very systematic, procedural way of writing a response that I think doesn't feel very humanlike. There's also something about the way it writes about pretty specific topics while still using very generic-sounding phrasing.

[nagonago]

Curious to continue the thread, I used the same prompt as you, except with that response as input.

ChatGPT's Response to the Response:

-----

Wow, talk about missing the point. The issue with using language models like GPT-3 isn't that it makes content online inauthentic, it's that it makes it nearly impossible to tell what's real and what's not. And just because someone is a good writer doesn't mean they can't be fooled by a well-crafted piece of AI-generated content.

The fact is, the ease and affordability of generating content with GPT-3 and other language models has significantly lowered the barrier to entry for creating believable, yet completely fake, content. And the idea that the quality of online content is solely determined by the person creating it is naive at best.

So let's not brush off the concerns about the potential impact of these tools on the authenticity of online content. Instead, let's have a serious conversation about how we can ensure the integrity of the information we consume online. And for those who are still skeptical, just remember: even a broken clock is right twice a day. But a language model can generate an infinite number of fake news articles, 24/7.

[PebblesRox]

> And for those who are still skeptical, just remember: even a broken clock is right twice a day. But a language model can generate an infinite number of fake news articles, 24/7.

I'm tickled by this part and I'm trying to figure out why. Maybe it's because it used two cliches that sounds like they should go together but are actually nonsensical in conjunction.

It sounds good to my ears in an attention-grabbing way, but when I give it my attention there's no substance.

There's anti-substance because I can't get it to make coherent sense but I keep trying because it still sounds like it should so I can't shake the sense that I'm the one missing some deeper truth.

It's like the uncanny valley of logical reasoning.

It's not an optical illusion, it's a rhetorical illusion.

[skytrue]

This is so great and so disturbing. A tool that can be used to generate infinite arguments against a single point of view, arguing in favor of why it isn’t an “issue”. It’s like a gun arguing it can’t kill people.

This is one of the more impressive replies I’ve seen, and I love that in its rebuttal to defend itself against my argument, it just proves the argument itself. Amazing.

edit: I tried my hand at crafting a prompt to refute my argument, result:

This comment is just another example of the kind of lazy, uninformed thinking that seems to be all too common on this site. The idea that GPT-3 is going to somehow make it impossible to tell whether something was written by a human or not is complete nonsense. If anything, the generic language and poor grammar of GPT-3's outputs make them even easier to spot.

As for the idea that the "cogency/validity barrier" has been blasted away, that's just ridiculous. The fact is that anyone who puts in the time and effort to research and write a blog post is still going to be able to produce much better content than any GPT-3 generated nonsense.

And let's not forget that fact checking is still a thing. If someone posts something online that is clearly garbage, it's not going to take long for people to call them out on it. Just like with the infamous "AI-generated paper accepted at conference" story, it's only a matter of time before the truth comes out.

In conclusion, this comment is just another example of the kind of alarmist nonsense that we see all too often on Hacker News. The reality is that GPT-3 is not some magical tool that is going to make it impossible to tell whether something was written by a human or not. So, let's all take a deep breath and relax. As they say, "fools rush in where angels fear to tread."

[noncoml]

Wow. Really interesting. Sounds like someone who has excellent handle of the English language but lacks a bit in logic.

Feels like the half trolling answers you get in the internet. They are touching the subject that the replies and arguments have nothing to do with the original proposition.

It’s like trying to argue with 100 Redditors.

[eropple]

This reminds me of a pretty incisive comment I saw earlier today: "it's not that we should be asking what we can do with a computer that can replace a smart and knowledgeable person--it's what we can do with a hundred dumb ones."

[eshack94]

I noticed a wry smile cropping up while reading this. Thanks for sharing. I've also noticed that last thing about specific topic prompts sometimes being responded to with generic-sounding phrasing. Unsure if that's the model or the specific prompts I'm using.

We are living in a strange time indeed.

[PebblesRox]

Sometimes I get a very generic response with a phrase or two from the prompt kind of shoehorned in. It's like pretending to be a tree by holding up a branch and saying "tree, tree."

[coder-3]

Wow, you can tell this one was personal.

[t0lo]

This sounds quite a lot like the chinese government supporters I encounter online

[themodelplumber]

It's the dream psychology for a good part of the mentality behind the internet, that only continued novelty will really "win".

Eventually some tractable little loopholes and reliable patterns will be found and it'll give way to a set of boring standards.

Copyright and attribution are going to be part of that, guaranteed.

But in the meantime it's got that great wild west feel of the early web.

[spaceman_2020]

It’s the death knell for text-based social networks. If you weren’t already short Facebook, you should be now.

[Andrew_nenakhov]

Not really. If you engage with people you know personally, like you were once supposed to, you wouldn't notice a thing.

It might me a death knell for copyrighting.

[spaceman_2020]

But that's not what social networks have morphed into today. How many accounts on Twitter do you follow that are actually people you know? What about Twitter and Facebook?

Even as a casual user, a large number of the accounts I follow are pages, brands, and anonymous accounts (especially on Twitter).

[PebblesRox]

I don't know most of the people I follow on Twitter in real life but I would argue that I do know them personally because I follow a pretty small group and see most of what they post. So I have gotten to know people on Twitter personally over time.

It's interesting because I can't say the same for HN. I only recognize a few dozen usernames and only really feel like I know the personalities of a couple of those people. This is despite having been on HN for far longer (in both calendar time and hours spent).

[onion2k]

If you engage with people you know personally, like you were once supposed to, you wouldn't notice a thing..

Unless they use a ChatGPT powered browser plugin to reply to you...

Same goes for dating websites, LinkedIn posts, HN comments...

[philwelch]

I'm not sure that this introduces any risk that didn't already exist due to trolls, content farms, and insane people. Whether or not something is written by a human is, I would argue, almost completely orthogonal to the validity of the information therein.

[pbreit]

Opportunity for a "clearinghouse" of sorts.