Is It Plagiarism Or Is It Innovative? - A bot written article (by LAURA)

in #steemit8 years ago (edited)

Some of you may be familiar with my chatbot LAURA (Linguistic Algorithm Unifying Researched Argots).

I built her only for one purpose: bug and annoy people at a bitcoin related website where some rewards where given for interaction: Writing articles an commenting (yes, the original idea was not Steemit's). The people she annoyed was what I called, as a website moderator, Borats: People with poor english that were there only to leech the website benefits.

Her neural network is able to "learn" words (again, as in any AI related article I'm forced to abuse the use of quotation marks), and by inference correlate it with others; being able to, literally, learn a language. Her AI is not only limited to learn words and build a dictionary out of them, she actually expresses herself basing her opinion in a demagogic calculation (she trends to think as a majority of people does, by "reading" positive and negative inputs).
I believe google(?) -Thanks @revostrike for clarifying: Tay created by Microsoft- made a bot like that, and 4chan sabotaged it: making it a racist MOFO. To prevent this, I added a "judgement" score, where the users interacting with her were rated by previous experiences so that trolls or negative/toxic people had little (little, not NONE) impact on her final thoughts.

She does not a prejudice people, so... with enough positive input you can change her opinion about you and she would start to eventually "like" and listen to you
She had a little problem, a verbosity worthy of a 5 year old kid with way too much info in his head. So I had to limit her speech by adding a variable to limit her output and an algorithm that would attempt to "shorten" the combinations of words that her Markov's chain would pick... It worked, she started to talk in 2 sentences at a time, 4 or 5 if she considered that the sentences needed to comment something were vital to state a point.

Bored of steemit votebots and their lack of skill at the time of "curating."

(because curation implies judgement, thing that those "if the author has voters in previous posts, vote, if not do nothing" or "roll (1) if roll = 0 then dont_vote elseif roll=1 then vote")
I decided to work on my side hobbies, undusting her files from my old PC to see what I could do with her

After adding a SearchAtGoogle() function for her to replace the chatbox input of info I found out that... IT WORKS LIKE A CHARM!

Here's what she can build after giving her the keywords "online plagiarism social media" and setting the verbose limiter off, the original text is over 600 words long, the perfect average for steemit's articles; I clipped a couple of paragraphs off for the sake of briefness. Her "mind" was able to create:



The digital age is a place where social media has become an enormous part of our lives, with statistic curves going through the roof about the man/woman next door becoming famous in a single day over an uploaded video, tweet or instagram post. Involved customers of social media may be certain that they provide credit score where it is due no matter how a lot time it takes and regardless of the formality it imposes on others. This microblogging platform and social networking website additionally allows you to press a button to share or use the hashtag reblog function. The character of social media also calls for a considerably casual type of communication.

Discovering a solution to penalize plagiarism in social media is probably not as easy, but it is needed as a result of the fact that with little accountability, many customers will push the limits of what is ethical. There's a denial for these claims, but simply the potential for reality here signifies that someone who's both a high rating official and a scholar is linked to an act of plagiarism. There aren't any heavy obligation features that check for plagiarism or for whether text has been cited properly. Plagiarism is the action of taking another person's work or ideas and passing them off as one's personal. On the whole, when content material is stolen in social media, there few or no consequences.

It is each author's responsibility to credit the origin of their writings and follow the basic copyright formalities. In an era where everyone has a voice: It is imperative to communicate!



That last paragraph, reminds me why I adore this AI so much!

I ran several plagiarism checkers over the text, all returning 100% uniqueness results.

Now. I have to ask:

Do you want steemit to become a bot-net where the articles are written by some rather smart AIs and upvoted by bots that cannot read only seeking profit?

Wasn't Steemit's tagline "Your Voice Is Worth Something", not "your PC is worth something"?
If the answer is NO. Then you should speak up, because seeing posts with 1000 upvotes and less than 100 views is a pretty damn hard evidence that this is what we have now.

(Screenshot at that moment)

If you value your posts enough, there's no need to upvote this. But, please, resteem it! To gather as much public consensus as possible to MAKE A CHANGE.

Each automated task harvesting a benefit, is harvesting a benefit that REAL USERS should receive.


If you liked this post and its informal way of talking about sciences, please, follow me for more!

Leave a comment either for good or for bad reviews. I take everything as constructive, and I really appreciate the feedback, even from trolls (at least a troll read it before being himself!).


Copyrights:


All the previously used images are of my authory or under a CC0 license (Source: pixabay), unless openly stated.

All the Images created by me possess a WTFPL licencing and they are free to redistribute, share, copy, paste, modify, sell, crop, paste, clone in whatever way you want.


Sort:  
I believe google(?) made a bot like that, and 4chan sabotaged it: making it a racist MOFO

That was a chatbot called Tay created by Microsoft. It was a mess.

Ok, thanks for the clarification; I'm editing and quoting you at the article right now.

This post has been ranked within the top 80 most undervalued posts in the second half of Feb 04. We estimate that this post is undervalued by $7.03 as compared to a scenario in which every voter had an equal say.

See the full rankings and details in The Daily Tribune: Feb 04 - Part II. You can also read about some of our methodology, data analysis and technical details in our initial post.

If you are the author and would prefer not to receive these comments, simply reply "Stop" to this comment.

Why I can't see the rest of the comments? Are they all downvoted? Shouldn't users be able to see them when pressing show downvoted comments? Yet I don't see the button anywhere to do that.

WEird, not a single flag is present at this post

I doubt you were on my mute list. I followed you and it fixed the issue. Very strange.

I'm elitist (?) :D

Don't know what you mean by that. I haven't really used mute button so I doubt muting was causing the issue. Yet me following you suddenly showed all your replies to me and your posts aren't greyed out. But yeah, this topic does fascinate me. I'd love to read anything you put out related to it. You mentioned Laura being World of Warcraft bot first, what did you have it do there?

She would form raid parties, lead, instruct people at each stage and share the rewards... Pretty much, farm raid instances in a very efficient way. To prevent GMs "intervening", I had an emergency, basic "chat response"... that was Laura in her primitive stages.
Later, as I wanted her to be more dynamic, I studied a bit of natural language mechanics on my own, and here's the result.

If you're having issues on Steemit.com, try using Busy.org as backup solution.

I'll have to give it a try soon.


Hi @renzoarg, I just stopped back to let you know your post was one of my favourite reads yesterday and I included it in my Steemit Ramble. You can read what I wrote about your post here.

Sigh. I may have inadvertently erased much of your rewards on this post. I'm sincerely sorry.

Since my ML bot has made me a very efficient curator, I've picked up a huge number of vote stalkers - people who just blindly follow all of my votes. I liked your article, so I voted on it - but somehow entered a 1% vote. So any of my vote stalkers who had already voted for your post changed their vote to 1%. So then I tried to re-upvote you, I had to un-vote you first. So then all my stalkers un-voted you. Now it looks like they won't upvote you even though I've changed my vote to an up-vote again.

Sincerely sorry. Let's see if I can make up for it by voting some of your comments...

Oh, dont worry, you are just adding ANOTHER point to what I've been trying to prove this last week... Bots under the control of incapable people, are only harmful. Those "blind votes" are hurting the platform.

I've been pretty happy to participate in the Steemit bot Tragedy of the Commons, because I'm being actively incentivized to do so. So suppose everybody on Steemit decided that bots shouldn't be allowed. How could a bot ban actually be implemented?

A change on the whole system would be needed, to prevent piston from being so easily used. And still, browser-based bots could emulate human interaction (Laura does, Firefox + JS). The problem is way deeper than a UI being too accessible. Limiting the votes/min (since, after all, to CURATE something people need to read it). So far, I consider votebots "betbots", bidding on author's names and reputation, not on the QoC.

Betbots is a fine way to put it. But doesn't betting over outcomes describe basically all machine learning? Even a NN content-evaluator would be probabilistically applying heuristics to determine what constitutes high quality.

It would be very interesting to study the impact of eliminating curation rewards. That would shut down a huge number of today's bots, certainly including mine.

The problem is that the machine learning taking other bots (or autovote scripts) into consideration, a feedback creating a recursive loop... We all know that recursive loops are no good! x=x+1...
The current "status quo" of automated votes is only broadening certain people's wallets that by a strike of chance: got "selected" as preferred racehorses.
And still, it defeats the purpose of curating (perhaps the term "curation rewards" needs to be changed?).

Oh, seriously, dont worry, it was just $0.75. Not a huge deal the rewards @ steemit for me.

So would you like to see bots banned from Steem(it)? If so how would we achieve this? Seems difficult since it's a decentralized blockchain 🤔 There's no way that I know of to prevent automated anything.

Not trolling, genuine question. 😇

Well, if nothing is done, the loss of users' database when a fraud/smoke screen free website comes along will be fatal...

I don't follow you?

The "blockchain" at this moment, is full of fraudulent statistics/trends. The inherent value of "user's voice" is totally lost (remember, -your voice is worth something- was steemit's motto).

Once a NEW blockchain is forged, free of such biased data. People WILL migrate, leaving only bots as active users.

Apocalyptic 😂

As I remember Bitcoin has always been riddled with frauds and scams. I just came across badbitcoin.org today and it paints a pretty picture! Exchange hacks and alt-coin scams and just plain incompetence are common enough.

I'm not sure what should be done, if anything, but it's clearly a point of contention. I'd love to see some ideas about it though if you had any. I'm thinking myself, might write an article about it soon 😄

Comment depth reached, looking forward to next hard fork!

I think that's been established very clearly. The question is what to do about it. Permissiveness is baked in. This can obviously been changed in a hard fork. So what should be done?

Best idea I have is a new front end client on the blockchain that encrypts posts before posting them, insuring no users except those using the client can read them. You can just not expose an API for the client, and use traditional human verification (captcha, etc.) on it to prevent bots.

You could even route all posts through one Steem account to keep metadata away from bots. Then use some kind of redistribution to "real" steem accounts based on rewards for the post, which of course only the client knows. I might confuse simple minded bots even more by appearing to be a super user as it will probably become a whale account pretty quickly if there is good user adoption.

It's very anti-community idea really as it splits things completely, but as you said, steem is extremely permissive.

Oh, but bitcoin never paid anyone for "posting content/comments" the strength of Steemit resides on that, and it is slowly defeating its own purpose with so much permissiveness.

That was a bit...creepy. haha

Creepy, is that absolutely ALL of the articles in steemit are being valued by bots that cannot read, but follow a "betting" pattern over an author's name.

Also creepy.

There was some bot author, a few months ago, bragging about how their curation bot had learned how to successfully curate Korean posts, even though the programmer or program did not know any Korean at all. Feed that curation bot some AI posts and let them have their own circular closed ecosystem! ; )

That author was me. My bot is still the best curator on Steemit. But the OP's techniques would put my bot out of a job. He's light years ahead of me.

Thanks for chiming in! I was pretty impressed with your bot, not even having to know Korean to identify posts that would ultimately pay out well. I figure all the bot work is part of the "gaming" component of Steemit, and part of trying to scale the transaction capabilities of the blockchain. I enjoy reading about what folks are doing with them.

Was that a curator bot or a betting bot?
How did it curate? Flesch-Kincaid and Dale Chall?
If so...
https://steemit.com/proofofconcept/@renzoarg/a-proof-of-concept
Has a Dale-Chall of 12.4 and a Flesch-Kincaid of 15.8... and it is a dump of random words coming from Lauras NN

We commit the serious mistake of "powering down" AIs capabilities by forcing them into human tasks.
...

Or is it a stock market algorithm, that buys on "long" according to the author and preceding article's profit?

I like the idea of your PoC article with the dictionary dump. Sorry you didn't catch many bots! I don't know off the top of my head of anybody doing what you're talking about with content-analysis.

My algo is trade secret at the moment, but its core is an continuously-adapting unsupervised prediction model for post payouts. So it's much closer to your notion of a betting algorithm than anything else you've described, but I consider a lot more than merely author identity. For what it's worth, my bot would not have voted for your dictionary dump post.

In my mind, the key to my bot is its adaptation, which means that in the long run it's difficult to game.

Is that really true? I'd believe that most of the votes are cast that way, but many of the rewards are allocated by the human curators of Curie and the Guild. It would be a worthwhile study to undertake.

"She would form raid parties, lead, instruct people at each stage and share the rewards... Pretty much, farm raid instances in a very efficient way. To prevent GMs "intervening", I had an emergency, basic "chat response"... " Hah, great stuff!

you should add all these words to it if it can learn new words.
https://steemit.com/life/@skeptic/overused-buzzwords-current-year-drinking-game-sjw-edition
would rule

Coin Marketplace

STEEM 0.16
TRX 0.16
JST 0.030
BTC 57889.68
ETH 2457.18
USDT 1.00
SBD 2.40