Warn HN: The AI That Plugs Your Product on Reddit in Conversations | Hacker News Reader

Warn HN: The AI That Plugs Your Product on Reddit in Conversations

replyguy.com

74 pointsA_D_E_P_T80 comments

Recent and related (see top comment):

Reddit is full of bots: thread reposted comment by comment, 10 months later - https://news.ycombinator.com/item?id=40211010 - April 2024 (272 comments)

Congratulations on being the first person to do something with AI that will directly make my own, personal life worse. You are a monster.

You should not be showing this as if you are proud on HN. You should hide it and whimper in shame.

This is the first for you?

> that will directly make my own, personal life worse.

Have you considered doing things that will make your life better? I use ReplyGuy to save myself like 30-60 hours monthly on each of my projects and it has made my life better.

Thanks buddy i agree with every word you said and will check this link out as well, thanks again! /s

At the expense of making everyone else's life slightly worse through bogus product recommendations and slowly degrading the quality of the medium you advertise in? Guys, have you considered taking from the commons? There's all free stuff there, it's benefited me personally.

> Guys, have you considered taking from the commons? There's all free stuff there, it's benefited me personally.

Classic Reddit joke: “Did you know that the ducks at the park are free? You can just take them!”

I actually had to ask myself whether this was written by a real person or by a bot like ReplyGuy. I've never felt the need to ask that before. I hate this.

I thought I was so obviously over-the-top mimicking the bot from the subject link that it would be obvious satire (and thus point out how easily something could be done on HN)… But I’m not so sure…

I had to go back and undownvote your first comment. You got me, at least.

I caught the downvote bat for replying with one of the replies from the "People love our replies, really" section from the TFA verbatim :)

Heh, I was so paranoid about the bot thing that I didn't even consider it could have been sarcasm. I even checked your last couple posts on HN after reading your comment just to see if your account had a history and if it was all shilling for products.

I'm convinced you're a real person. Sorry for not catching on to the sarcasm.

You sound like a mechanic who dumps his used motor oil down a storm drain to avoid the expense of proper disposal. Yeah, it's good for you, but it pollutes the environment for everyone.

You are polluting the digital commons. You should be ashamed of yourself.

The user who posted this is perfectly legit, and not the person who created the site. As their follow-up comment makes clear -- they weren't doing this to promote the site, but to warn others of its existence:

https://news.ycombinator.com/item?id=40236860

As such the post is unfortunately all too relevant, and shouldn't have been flagged.

Yo, I didn't invent it. I'm trying to warn others about it.

Pretty sure this has been happening for a long time now. Before AI/automation, they would just hire people in developing countries to manually promote products on the platform. You can take some steps to mitigate this: for example clicking on the username of people recommending products and seeing if they promote the products more than once or have a history of being active on other unrelated subs. Although they also manipulate that sometimes (purchasing old/actual user accounts), but there's not much else you can do at this point. Even the old word-of-mouth method is more unreliable these days because people just echo what they saw on social media.

Stumbled across this earlier today.

I think it is becoming quite clear that Reddit is the most astroturfed place on the internet -- gaming Reddit for positive feedback and reviews is a lot easier and less painful than trying to game Google's results directly -- and there's an entire industry building up around it. Something to think about if you add "site:reddit" to product searches.

I think it's also pretty much impossible to avoid this.

As soon as something has enough trustworthiness and user base for product recommendations. It has value to the advertisers and they will "mine* out that value until it's exhausted.

Reddit has always been heavily astroturfed (cough, Eglin Airforce Base), but since the mod blowup last year it almost feels like the bots and astroturfers are outnumbering the real users. The site has become unusable.

As hard as it would be to enforce, I wonder whether this runs afoul the FTC rules regarding advertising/disclosure. It does seem to me like this qualifies as a “Robo-influencer” and thus should adhere to the guidelines the FTC put out [0].

[0] https://www.ftc.gov/system/files/documents/plain-language/10...

The link to the "Stealth Marketing" site in the site footer looks like they don't care about adhering to any guidelines...which I guess should make the FTC's case clear-cut!

Out of the links in the footer:

- Painkiller Ideas and Replyguy both use Azure DNS and Microsoft hosting which means the footer probably contains the creators' other projects

- while stealth.marketing is hosted very differently, it uses partially the same and partially very similar CSS

- stealth.marketing says it's "powered by reddit" with the reddit logo which I'm sure the Reddit PR department will absolutely hate when finding out, given they ruin the platform (probably not even using the API)

Looks like there's an easy way to make this all collapse

Yea this is gross. Even the examples they provide are gross. Someone struggling with debt, and a some robot gives them the impression that there's a person that cares and, by the way, you should spend your money on this service. Ewww.

> We are currently adding support for LinkedIn, Tik Tok, Hacker News, and other social networks.

Yuck

They are gonna get sued if they try this on LinkedIn or Tiktok for sure. Circumventing those captchas (everyone who's tried scraping Tiktok before knows what I'm talking about, even in headed environments) is almost certainly illegal.

lol no

Oh, absolutely. Never heard of the LinkedIn scraping lawsuit before? They are totally willing to throw their lawyers at you for using the platform in a way it's not intended to be used in

Curious how this person is getting away with it https://tikapi.io/

An AI that makes the internet even worse than it is! Congrats to the team.

More importantly I see this as a band aid for underperforming products. The startup space struggles with this already, but instead of focusing on the product deficiencies and building something people want now we just enable some SaaS that pretends to be a human spewing marketing masquerading as a legitimate human's perspective. Talk about a bottom of the barrel approach. The Internet is going to need to go insular and back to niche forums that are under lock and key from this toxicity.

All this is going to do is destroy one of the last remaining ways of getting an unbiased review: word of mouth. It won't make people buy more {PRODUCT}, it will just make people suspicious of every recommendation they read.

Maybe companies should focus on actually producing products that are good enough that they don't need to be covertly advertised to via CIA level psyops.

This entire thing makes my blood boil.

Slow though the FTC is I imagine that a bare minimum requirement for products like this will be disclosure that they've been written by an AI bot

Which will (a) Give consumers proper disclosure and (b) Give platforms the opportunity to filter or address as they like

Prior enforcement engagements like the CAN SPAM act etc didn't eliminate bad actors, but massively curtailed legitimate business participation in these areas

What's stopping them from just running the company outside the FTC's jurisdiction? It's not like the U.S. is the only place with internet access.

Many countries regulate fraudulent advertising, and the person _paying_ for it would generally be liable (along with the person providing it, usually).

I wonder if this is anything like the YouTube comment spam that appears under every finance-related video.

There are always entire comment chains talking about how important it is to invest wisely and then plugging some random person's name.

The only way this ends is with either requiring strict identity verification for posting, or giving up on comments altogether.

Their "Trusted by these businesses" list is ironically a list of businesses not to trust.

AI posing as human customers to promote your product on the web is fraud, change my mind.

If you've been on Reddit for a long time, IMHO, it was obvious that the front page was effectively taken over at least a decade ago by shilling and rigging (not counting recreational karma gaming that was going on forever before that).

And surely in recent years people have been doing things like this ReplyGuy, especially with LLMs and with lesser hacks before that, and more secretly.

What ReplyGuy gives us, with their apt roofie predator logo, is a face we can put to one of them.

You're a horrible person. Your mother should be crying. You're the vanguard of a trend that will damage the mental health of millions or billions, especially if taken at all farther.

Or maybe it kills social media entirely and ushers back in a world of in-person human interaction.

Probably not, but we can hope.

Their "People love our replies" section shows a bunch of RandomWord-1234 accounts giving positive feedback. I wonder if those are included with the service.

I believe those are the default Reddit usernames (if you don’t manually pick one when signing up, you get one like that).

Back in the day you would read something on the internet and wonder if it was a cat that was typing it, or a human. Now people feel offended it might be an AI typing it directed by a cat or a human

http://i1.kym-cdn.com/photos/images/facebook/000/427/569/bfa...

This should be illegal.

I'm guessing it's against Reddit's TOS but good luck to them trying to enforce it.

I mean, this will directly hurt Reddit's credibility and as such they will move to change things as the stock price will ...

Okay, sorry there, I couldn't even finish that. Laughing too hard.

Reddit.

Credibility.

I know, I know, too much for me too.

Profoundly obnoxious of course. It'll take some time, but hopefully a mix of manual flagging and automated countermeasures will be sufficient to quash this onslaught.

Meanwhile, as to the visionary genius behind this thing:

Experimental urbanist Scott Fitsimones shares how these mission-driven, blockchain-governed, collectively owned organizations could increase the speed and efficiency of building cities (among many other applications) -- all while pooling decision-making power in a radically collaborative way. Hear about how he started a "crypto co-op" that bought 40 acres of land in Wyoming and learn more about the potential for DAOs to get things done in the future.

https://www.ted.com/talks/scott_fitsimones_could_a_dao_build...

https://theresanaiforthat.com/u/scottfitsimones/

And the carefully chosen product name: https://en.wikipedia.org/wiki/Reply_guy

So what's the move from here? Back to consumer reports?

Pretty much. Trust has always been a problem, especially on the internet, but people like this guy are going to simply ruin the public, free internet and destroy the very idea of crowd source.

We are going to resurrect the elite class of editors so we can ever find out any information that is not totally corrupt.

This idea is interesting. It’s nothing new; there are already people hired to plug their tools or respond to complaints, even here on HN. This could make a “community manager” a lot more affordable. For business that is nice.

As a user I fear the future, but I don’t see it as much different from a human doing the same thing. It will just get a lot more common.

God do the bots really need to sound like such complete tools on top of being what they are?

On one hand I agree that this should be flagged and taken off homepage (as it is right now), but. Otoh more people need to be aware of such scams and simply not trust reddit, hn, etc as good source of information anymore.

Pricing is interesting: They have a free tier that provides 2 replies/month, next tier is $10/month for 20 replies/month. This is probably still more cost effective than replies done by humans though I understand that LLM API access can get pretty expensive at the moment, what models are they using?

Other than that it seems pretty cool, and ReplyGuy is a perfect name lol.

> We are currently adding support for Hacker News

They could be any one of us!

You are helping to ruin society , not that you care

Nobody wants this.

There are plenty of unscrupulous capitalists who will love this. Unscrupulous capitalists are people too.

I disagree (that unscrupulous capitalists are people, they are monsters)

It's weird to try to imagine why he thought showing it here would get anything other than hate. It's like saying he found a way to smear poop on everyone's doorknobs.

Something like "There's no such thing as bad publicity" I guess.

They're on the frontpage so it's definitely worth it.

OP posted elsewhere that he's not affiliated, FYI

Lots of people want this, just not us. We're not the customer. We're the product.

I'm sure some entrepreneurs would want that to sell products.

What if it's just a bunch of humans copying pasting stuffs? Doesn't seem to require a lot of replies so far.

"What accounts send the replies? We send the replies from our pool of high quality Twitter and Reddit accounts."

I wonder what they could mean by "high quality" here. And I'm guessing/hoping that these accounts won't stay "high quality" for long...

Omg PLEASE PLEASE do not do this. Do not piss in the well. Reddit needs to ban anyone who uses this.

This objectively makes the world a worse place. I challenge you to come up with any ethical angle that supports your work on this product.

If you believe the purpose of an economy is to meet peoples' needs, then it helps connect users that are openly looking for a tool to help them with projects that will help them with their need/problem/request, which is a win. Alternatively, if it really does make the world a worse place, then it incentivises governments to prevent their citizens wasting time looking at marketing online so they can get their act together and ban online advertising already, which is... a win. Hey, I'm just accepting your challenge.

when I saw gpt-3 in 2020 I figured this would happen. I called it Commentdämmerung, the twilight of the comment section.

somewhat legible explanation at the time: https://simonsarris.com/p/commentdammerung

You've created a disgusting thing that makes everyone's lives worse. And you'll justify it to yourself with the old "if I didn't do it someone else would have". Despicable.

This is tragedy of the commons

There's already a few 'this is a bad thing to exist on the Internet' comments, so I'll say something else.

The four example comments feel a bit off, because while they are tailored to the informational context, they don't fit into the conversational context. They all start off with an introductory sentence that restates contextual information that's already present in the discussion thread.

> To launch the chrome browser when you click the text, ...

If you're reading the comments, you already know what question this is an answer to - wasting people's time with this high school essay nonsense is going to trip bullshit-o-meters.

> Ireland, like any other country, has its fair share of crime ...

No one starts a Reddit comment with this sort of equivocation unless they're trying to head off an anticipated downvote storm. Generic travel safety advice certainly doesn't need this.

> It's understandable that you're feeling anxious and overwhelmed with the debt and job loss.

This one does a lot of waffling, which people don't really do on Reddit unless they're talking about something they're passionate about. Just give the useful advice and move on.

Overall, they have a lot of the padding-the-word-count style that characterises LLMs that were trained on blogspam. Redditors tend to be both more direct and more personal in their writing styles. The LLM isn't so bad that it'll be immediately identified as AI, but it will come off as subtly ingenuine.

Everything I've ever seen written by LLMs reads like they're a high school kid trying to pad an essay. I'm not surprised at all this does the same.

Does this break the OpenAI terms of service? Or whatever other API they query?

So that’s why the Butlerian Jihad started. Now we know, thanks.

Tweets from @ReplyGuyApp on Twitter.com [1] about a month ago:

> If Hacker News hates your product, that's what I call product market fit.

> When Dropbox launched, they got tons of comments like this:

> [Screenshot of a HN comment - you know the one]

> Hackers can build things themselves and therefore underestimate the power of productizing something. Our launch was met with the same skepticism.

I guess we're all underestimating the power of productizing… spam bots?

https://twitter.com/ReplyGuyApp/status/1773132507153654123

TBH this has nothing to do with hacking, well mental hacking maybe.

The original hacker did sell stuffs too but they mostly follow the path of introducing computing and products of computing to the mass market.

This one? It's just an ads grab.

Ugh. Frankly the sooner this bubble bursts the better at this point.

Good luck getting that to work consistently on LI.

I was upset about it at the time, but last year's round of hostility to users that caused me to delete my reddit/twitter accounts when 3rd party clients were lost was a blessing.

Reply on news.ycombinator.com