I DO NOT support AI generated text as Hive post

(78)inOCD•

Reblog

I DO NOT support AI generated text as Hive post

Lately there have been a proliferation of Artificial Interlligence generated text in hive. I am actually surprised why it took so long for hive farmers to catch on! Since at hive we get rewarded for text content as blog, and the fact we are relatively small (anything and everything gets voted, some to obnoxious amounts), it is rather easy to generate AI content at hive day in and day out like clockwork. However, I must say AI generated content is NOT original content. Therefore, I will personally try and not support such content. I will also personally try and remove rewards from such content.

Source

Due to the decentralized nature of hive, I have no authority of telling what hive can or can not do. It is upto the community to decide. However, I can make my thinking visible and I can also request and lobby for the cause to other hive stakeholder so that they do not support AI generated content.

How do we detect

Often, a lot of people who are leaning towards AI generated content are not native English speakers. In that case it is rather easy to figure out the unsual usage of certain words. This is simply because AI learns and it is very good, but it is still biased towards the input.

Community does have a couple of AI detection tools are are working behind the scenes. I am not going to tell what are they and who are behind it. I will only tell you that they are rather good, and have a very high success rate of detection.

My personal views are simple. If someone is smart enought to resort to AI generated content and posting for rewards, all they care for is money. In that case all they deserve is downvote. That said, there are certain gray area where I have seen people use AI generated art, but original writing to support that art and produce excellent content. I am okay with that. Again this is my personal opinion, and I am making it abundantly clear.

I am NOT going to discuss why I am doing it, because that discussion will not go anywhere. I am just telling you this is what I am going to do. You can do research on why it is not acceptable.

Embedded Image

·inOCD·by

(78)

$56.30

Sort:

(68)

Yeah, 100% behind that. A couple of thoughts of which I'm not so sure.

What do you think of smart curation bots who leave an AI comment instant of what we are used to right now with !LOL or !BEER

What to do with AI writing tools, I have a strong idea that those will be as comments as word correction pretty soonish and they will phrase/rephrase for us. Shouldn't we endorse those to increase the level of written content?

AI Pictures & AI Art seem uncopyrightable as intellectual property. Now people claim they modified the left cheek and now it's their original art, well I find it difficult to argue, but of course, The Alien Art Hive Community is doing that for years already. Any thoughts about those?

My last thought would be, that I wanted to see an AI Dev Community popping up on Hive for a long time already. Most of that ppl aren't Software developers but rather run a rented server or have a potent system to play with and add some scripts on a shell. It should be fine for them to share their progress under the correct taggings.

AI-enhanced or completely AI written Content has been on the Chain for more than a year at least. When I started calling out accounts, none seemed to care. In my opinion, the Downvote and Reputation system "as a system" has miserable failed us, but there will be AI Content Attacks on Hive and we might want to get in a position to see them coming and use the tools we have.

$1.36

9 votes

(78)

lots of questions here, and if I do justice to all of them properly that will be several posts, but let me try. This is my personal opinion obviously:

Curation bot comments: I don't like them. I think they are spammy and they make my contents comment section ugly. That said, I don't vote them. Since hive is decentralized I can't control them either. So I have decided not to care.
I am against AI to assist text writing in general as they are not original
Art is a difficult subject. As you have said AI art is "uncopyrightable as intellectual property". This has been decided. So people can use it, even without citation, and there is nothing wrong or unlawful about it. Again, to be it is all about reward, do I vote such a post that only has AI art in it. I don't. If it is mixed with some original content, depending on content I might. I often do.
I am aware of AI developers. At hive, as a content generation tool I personally do not support them

$0.19

2 votes

(68)

I will not further inquire, since you've been forthcoming and answered all of my questions.

Ha, we're at the front line, aren't we? We're going to find out first if and how it's possible to fend off AI 'attacks'.

$0.00

1 vote

(74)

I don't like them. I think they are spammy and they make my contents comment section ugly.

My opinion is:

It let's me track if someone is misusing curation project that are under my management. We are building a better alternative like tipu but till then spamming comments looks like the option to track. (Your post has been curated by "curator name")

Example:

25 people can vote from an account then how to track who voted what? Who broke the set standards?

Also, we already have a solution to "click to reveal" for comment curations. Don't seem like high priority. Personally IU, Discovery-it etc comments are muted by me.

$0.00

(81)

Shouldn't we endorse those to increase the level of written content?

Hive is about people, and stories - not how good they are at grammar.

When I started calling out accounts, none seemed to care.

I do! And would have, had I seen it. IN my community, I'm quick to mute anyone who isn't writing original content.

$0.03

8 votes

(68)

Hive is Blockchain, the Hive community is about a bunch of things. Original Content OCD Style is the most significant part looking back? Not sure.

The most significant parts are the communities themselves and Splinterlands so far. Meanwhile, the communities are 98% of the time in Discord.

Just giving you a reality check here. We're all part of these ideas surrounding Hive. People and their stories shared in an old-school blog format are one of those.

I'm open to arguments, representing my very subjective pov.

$0.00

(81)

Meanwhile, the communities are 98% of the time in Discord.

Plenty don't do Discord. Plenty. Or, they use it minimally to attend to admin matters. Used to use I all the time and run a massive community there. I don't anymore and avoid it like plague. Many believe Discord is a waste of time and we should be on chain. Threads is a good part of making that happen. I hope.

Interesting stat, 98 percent. Where did you pull that from?

Speak to any curator on Hive, who form a massive part of how content is supported and rewarded here, and they'll tell you original content is everything. Gaming and other ways to earn on the Blockchain are separate from blogging. Of course OC blogging is only one aspect of HIVE, but that's the part curators of blogging content are concerned with here.

I'm not sure what you mean otherwise. What are you trying to say here?

I'm not anti AI. Just anti it being abused to milk the system.

$0.03

8 votes

(68)

What I'm saying is:

The Hive Blockchain is not a webapp, what web2 access point are you using for these comments? PeakD?

Here's the latest post of dalz: https://peakd.com/hive/@dalz.shorts/comments-on-hive-are-up-durning-a-bear-market That's a lot of comments, I can't do the footwork and compare it to all the Discord channels in action. It's very unlikely that 98% is anywhere close. For some very active Hive Discord users, it's closer to 99.9% and for people such as you, it's the other way around.

Fair is fair, do Hivers exchange 35 million Messages per Month in Discord to create the 98% ratio? That means 13.5 Messages per Second across all the Discord Channels. Wow, well, it's not impossible but highly unlikely, very very unlikely.

$0.00

(81)

Oh so you are saying a lot of a activity isn't actually blogging? Sorry, having trouble understanding what your point is. Yes, that's true of course.

Yes, Peakd.

Sorry I'm struggling to see how this relates to reward pool abuse by using AI to write majority of blog. My bad.

$0.02

5 votes

(68)

Yeah exactly. Money can always be broken down to trust. Hive Value = Trust in Hive. So the way to interact via Blog was obviously the first entry point. But there's more to it.

It's obvious to me that if the usage of AI helps users to strengthen their bonds and interactivity, the community will become stronger and the Hive Coin Value will therefore rise. That's me and I is only subjective, but the argument against AI-improved content feels to me like a Disney Movie - but the animals can't speak to each other. A chat room full of people from across the world and the admin turns out the auto-translate to English, because of maybe valid reasons. Instagram deleting all Filters for Human Portraits, because it's inauthentic. Do we sacrifice to the holy lamp or are we still cooking?

I can't answer that yet, also I don't have to, because I'm not a community leader. Different Communities will take different paths and in two years, we will be able to compare by looking at objective truth. That's part of the human experience, the Future is speculation, and the present is a very thin layer tainted with subjectiveness, the truth can only be found in the past.

$0.00

(68)

The abuse part is not what has to be discussed, the very definition of abuse has to be adjusted. From word correction by Grammarly to AI rephrasing your written Blogpost down to getting full posts from Bing AI and just rearranging for personal preferences. What is abusive and how can we detect and fight it?

Do we all want to get rid of malicious actors who try to abuse us? Hell yeah.

$0.00

(68)

Thanks for this interaction btw, love it.

$0.00

(71)

lolztoken.com

What do you call an airline full of bald people?
Receding airlines.
_{Credit: reddit}
@azircon, I sent you an $LOLZ on behalf of @manniman

(1/1)
Use the !LOL or !LOLZ command to share a joke and an $LOLZ

$0.00

(80)

I'm with you on this. It's crap and if I see it I will remove as much as I can from it. Sheer lazy farming

$1.11

3 votes

(78)

It is becoming a big problem. I suspect some rather prominent people are using some variation of this for a while now.

$0.00

(68)

Yap!

$0.00

1 vote

(80)

Yes, I do too. What I think they do is put a prompt in then tweak the text of the results so it looks slightly less AI.

It is still quite obvious to see as these people are invariably lazy as well as not hugely bright :0D

$0.00

2 votes

(78)

I am willing to adjust rewards just on the basis of suspicion, because the posts I am thinking about are grossly over-rewarded anyways. I don't need any other reason usually.

$0.22

2 votes

(68)

If a post is not very formatted but perfectly written and makes very few personal statements, that is a red flag for sure.

Makes me wonder more and more if we as a community should have a digital ID to verify us. No KYC or something like that. A novel digital Hive-based ID to create deeper trust.

$0.00

1 vote

(80)

Quite rightly too, it's not as if they would admit it. Or if they did it would be worth done day caveat that they used it as a tool and only took bits of it or some such tosh.

$0.00

(81)

I"ve heard of 'big name accounts' doing this, and I'd sure like to know who - I'm soooo curious!!!!!!

$0.00

(80)

Hahaha, they are clever but you can spot em if you go a looking hard enough. ;O)

$0.00

(69)

Although I do agree that AI-generated text being passed off as original content should not be rewarded (on Hive or elsewhere), I fear that it’s going to be exceedingly difficult to stop.

As a university professor, I’m convinced that there’s no way we can continue relying on essays as a way to gauge student learning and comprehension.

I’ve taken steps to make it harder for my students to rely on AI to write their essays, but it’s an uphill battle that’s only going to get harder.

Even with OpenAI’s classifier, all the students need to do is revise the essay enough to fool the classifier, and eventually other AI tools will do that for them.

$0.43

1 vote

(78)

We will do what we can. Just like what you are doing for your students. We will make it difficult

$0.66

1 vote

(74)

I was talking with hubby about this just yesterday. At some point the only way to be sure in schools may be to go back to writing by hand. It seems ironic in many ways, yet it's only really a generation ago that we all hand wrote essays.

$0.58

1 vote

(69)

I actually had my first encounter last week with a student using ChatGPT to write assigned essays.

Quite frankly, I noticed that the essays were 'too good'. When confronted, the student initially denied using any generative-text tools, but has since come clean.

As I stated in a post about ChatGPT several weeks ago, educators are going to have to abandon essays as a means of measuring student engagement and comprehension (unless the essays are written 'by hand' or proctored).

$0.36

1 vote

(83)

I agree 100%. I don't like AI writing or art...lazy, uncreative, unimaginative and farmy.

$0.16

4 votes

(78)

We got to keep at it. There are multiple high rep individuals consistently doing this for reward farming.

$0.00

(83)

Yep, it's reasonably easy to spot. I don't think any of us should be accepting of it though, it's misrepresentation.

$0.00

(81)

We need an 'Ai Watchers' comment on them, like Hive Watchers. I'd really like to know who these accounts are. Once, okay. After a warnin g - nooooo.

$0.00

(74)

I have never been a fan of the downvote function, but for AI generated work ... absolutely. To the language barrier; there are translate apps that allow people to communicate across languages using human generated text.

$0.04

4 votes

(78)

I am a big fan of downvote. This is the only tool we have to restrict the abuse of the reward pool

$0.00

(80)

Smart move otherwise it's going to turn into a crazy spam fest of junk with no one interacting sadly.

As "smart" as people think ChatGPT is it's actully super stupid it's just good at pulling older content from 2021 or later and rehashing it out with filter words etc. The only thing impressive about it is the speed at which it can do it. Besides that it's dumb AF

$0.04

1 vote

(75)

It's good that it has been introduced. Now, hopefully, people will think twice before casting their blind vote. It would've been a better practice to get to know the author/blogger a bit better first anyway IMO.

$0.04

1 vote

(78)

As you know very well, there are many AI text (especially low quality AI texts) floating around hive for a while. It is our collective duty to identify them and have a direct conversation.

$0.00

(75)

Well, I don't know if it is many, but I have seen some that made me wonder if this was actually written by the person. The contrast between person and story was too big -from my gut feeling-. But then again, and this is what makes it difficult to accuse someone for copy/pasting AI generated text, if you're running a business, you probably have been thinking about creating (or you already have) a tone of voice for your business. For example, my blog posts on @xposed, they are in a total different writing style than my personal stuff, but when I write about more "important" things, I can spend a lot of time on the wording. And when I start talking, I probably trip over a few words because my mouth can't keep up.

I have been testing a AI-text generator detector out. Surprisingly, its detection is not as accurate as you may think, which sucks ass, as it can point fingers in the wrong direction (and the irony? it can be used to create text to fly under the radar). Just more proof that AI is not to be trusted without DYOR.

We're dealing with people whose native language isn't English (mine isn't either for example), it's probably pretty easy for people to use AI generated text to translate something from their language in another, right? The stupid thing is, copyright can only be given to -humans- and not AI, because it isn't human (lol).

One way to kill users who publish AI generated texts, is to do a duty as curators: get to know the author first before casting big votes. It takes a long time to create an authentic personality on Hive. AI and bots don't stand a chance against authentic behavior. And if we're 100% sure something is AI generated, nuke them.

$0.04

1 vote

(74)

get to know the author first before casting big votes.

That's what I've been thinking myself for some time now. Unfortunately, it's not very encouraging to newcomers. On one hand we want to encourage people to stick around, but because of so many bad players it's hard to trust anyone new...

$0.05

1 vote

(78)

We do what we can....there will always be things which is uncertain and on the cusp.......but there are plenty of things that are 100% certain. We nuke them.

$0.02

1 vote

(75)

Found one.

https://peakd.com/@tornad

$0.00

(78)

Nuked

$0.00