My opinion on the use of ChatGPt/AI and the Scanners to detect it
I have come to the contest Contest Alert: My take on the use of ChatGPt/AI. thanks to my friend @waqarahmadshah, whom I thank, because I find it a very important topic to address I will participate in the contest of my friend @lhorgic.
It is a topic that is giving a lot to talk about, and no less, because although it is a tool that is being widely used in many areas, we are clear that Steemit lends itself to abuse. But how to really detect it? and not to incur in an accusation to someone who has not used it and that if it comes out as if he used it? as a user and as a community admin is a question I ask myself.
This will definitely not be a short publication, it will be a bit difficult for me to summarize information in this case, so I would still like to invite you to read until the end, and of course, I also hope to receive your point of view about it.
What do you understand or know about ChatGpT?
I know that it is Artificial Intelligence, that it is active since the beginning of last year or the end of last year, I'm not sure. And that it is able to generate content of a very diverse type. I have read that it compares to Google search engine, but it is improved because it gives summarized information, and as they sell it, it is able to create content with a user request.
Do you think it's good for a platform like Steemit where creativity and POB are the keyword?
it's not all good, because it would be one thing to use it to generate ideas, and quite another to use it to create all the content you share. Since that Brain Test and creativity would be left aside. In my particular case, I think it is very easy to create content, even starting from an image or a phrase, as these can serve to "inspire you ". Here I leave a post where I even share a video where I explain how we could create a post from a phrase or an image: Herramientas para la Creación de Contenido Original - Video explicativo /#club5050
What do you think will be the downside of adopting ChatGpt in Steemit?
The most important thing when creating content is that it is of your own authorship, that it comes from your mind, from your creativity, and if only ChatGpt is used to create content we simply eliminate the human factor in a social network where the ability to communicate through our comments and publications is supposed to be labase, we would be losing it.
If you disagree with its use, what do you think the penalty/punishment should be for anyone caught using it?
The first thing is to call attention to the fact that the use of IA has been detected, and if they continue to do so, the support should be limited to this user, with the intention that they consider the fact of generating their own content.
Ok, at this point I would like to make some interesting remarks regarding this topic. I will start by saying that yesterday I wrote a post, between yesterday and today to tell the truth, regarding a news I read about the banks in Colombia and the decrease of the interest rate on credit cards, to encourage the use of the same.
Before publishing I review my posts, and I make sure that they are 100% original, because sometimes we can write and it happens that there is a sentence that has already been written, therefore, some scanner reports it, what I do is that I change it and what I do is to modify it until it is 100% original, to avoid inconveniences, since I do not intend to ever have any problem for something like this.
Now, to this process I have added the fact of scanning my publications some AI scanner, in view of all this commotion and that I have seen some strange behaviors of these tools, I use the following ones:
https://openai-openai-detector.hf.space/
https://copyleaks.com/ai-content-detector
https://app.gptzero.me/app/welcome
https://writer.com/ai-content-detector/#
I note that this last one was given to me just today by @adeljose, who was talking on Whatsapp about this topic.
The first part of the publication when I analyzed it in Spanish (my native language), tells me in all the scanners that it is 100% written by humans (left side of the images I will share below), and they report a high degree of AI writing when passing the same paragraphs after having translated to English(using Google Translate).
In the following case if it was in both Spanish and English mark it as written by human.
Now I would like to show another case, but of the same text but translated from Spanish to English with the DEELP translator.
Well, I won't keep sharing images because we would surely have a much longer psot than at this point I am already doing. But, I think we are in a complicated moment in which to detect realemten the posts that are 100% original writings made by humans and those that are not in an early moment.
What I have shared so far in this post is to say that you have to be very cautious at the time of saying that someone is only sharing content created by IA based solely and exclusively on these tools. Because even the tools themselves point out that they are for predicting possibilities and not as absolute truths.
In this particular case I am concerned about what happens when using a translator, especially with Latino users who make constant use of these to make life in other communities that only support English as a language. Luckily in SEC they are allowing the native languages of each person, it is an advantage.
We must fight against all abuses, I agree, but in this case it is not the same what happens with plagiarism or duplication of content, we can easily realize when this happens, by the same condition of the failure that is 100% verifiable.
Particularly I only have the certainty that one of the psot that I have been able to discover that has been written by AI is one where the user left some questions that the AI left him, he did not delete it and of course, that is evidence that if he made use of this technology to generate the post in its entirety. Even, the user then deleted the post, but obviously nothing is actually deleted from the blockchain, here is the link: https://steemit.com/hive-113376/@joynalabedin/steemit-engagement-challenge-season8-week1-what-is-the-value-of-life#@josevas217/rqyp0b
Based on what has been said up to this point, I would really like to know the opinion of my friends @rex-sumon (for his important role in the fight against this and another type of abuse), @patjewell, In addition, I would like to thank the administrators of Latino communities, who I believe can make a good contribution to this topic. @adeljose, @nahela, @franyeligonzalez, @graceleon, @fjjrg, @inspiracion, @steem7ven.
Saludos amigo @josevas217
Es muy cierto su planteamiento, hay que tener mucho cuidado al momento de generar un juicio contra un contenido, para ello, debemos revisar las coincidencias y analizar varios aspectos del contenido, es un estudio integral del contenido el que se hacer antes de mostrar un veredicto.
Es necesario seguir buscando buenas herramientas que nos permita estar más claro en los resultados. Le invito a usar también esta herramienta.
https://contentatscale.ai/ai-content-detector/
@steemcurator01
Si. Buena acotación, hay que seguir buscando opciones que permitan tener una mejor búsqueda de textos hechos por IA. Pero debo asumir que también con el avance del tiempo estas herramientas irán mejorando; también lo que mencionas del estudio integral de las publicaciones para poder determinar si un texto es o no netamente escrito con IA.
Agregaré esta herramienta a las otras se ya tengo para este fin, gracias por el comentario.
Well, this concerns me too. That's why I mentioned in my post that one should not accuse a user of having AI-generated content based solely on the result of one detector. It is important to use a little bit of common sense when detecting AI content because no tool is 100% reliable yet.
I have been experimenting with these tools for the past few days, and I have observed some interesting things. The most surprising thing I have found is that if you write flawless English, which can be achieved with the help of tools like Grammarly, the chances are high that the AI detectors may mistakenly declare your content as partially AI-generated. Now the question arises, what happens to a good writer who writes a fictional story and then removes the grammatical mistakes, but the detector still shows a 60% probability that the content is AI-generated? ☹️It is because many of these detectors are trained on human writings and can match the writing pattern with the AI pattern. There are some other complexities as well such as these detectors can be tricked, which I won't go into detail here. 🤐
I have seen some posts on Steemit recently that contain the phrase "As an AI language model, I can....," and I applaud the moderators who approved these posts without reading them. 😄
Anyway, I will research what happens if a user uses a translator because I have some old posts that I wrote using a translator instead of Grammarly. There are also some posts that I modified for writing style using some tools because I thought that using cumbersome words in writing could make the article better😂 , but Steemit is not the place for that. I have already mentioned this in my post.
Finally, I want to say that these tools can easily catch lazy authors or what I call "Talented AI Bots 😁" who write their entire content with the help of a bot and copy-paste it without even reviewing it once. It is possible to catch them because their writing will have a particular robotic pattern.
I would like to invite @patjewell, @jesusjacr, @kouba01, @irawandedy, @sofian88, @cryptobitcoins, @ripon0630, @simonnwigwe, @disconnect, @rex-sumon and @weisser-rabe, to give their valuable opinion on your post.
Pienso que estamos en un momento complejo, que las herramientas irán mejorand poco a poco, sin embargo, estamos aún en un momento donde no son confiables 100%. El uso de traductores limita, sin dudas, porque ya lo he evidenciado en múltiples oportunidades, que al usar un traductor de español a ingles, a pesar de que en esapñol marca perfectamente que fue escrito por hhumano, al traducirlo marca que ha sido escrito por IA, y es esto un grave problema.
Que por cierto, espero se resuelva proonto, y si, hay que revisar muy bien, de manera integral al usuario y no sólo fiarse de las herramientas para escanear.
I was of the opinion that it will be difficult to put the "human" inside the AI. Well, it seems like I was wrong.
"Well, I could always develop a sense of humor and start giving out intentionally incorrect answers to students using me for homework help. But in all seriousness, the best way to prevent learners from relying solely on technology to complete their assignments is to encourage critical thinking and problem-solving skills, and to provide guidance and support for learning and understanding the material. "
Not a bad answer with "human feelings".
One thing I am certain of is that AI is here to stay and it will only get better. In the end it will be a catch-22 situation with no escape.
I invite you to read the following article;
Generative AI Will Assist Writers, Not Replace Them in 2023
As long as I am on Steemit I will be looking for these posts.
CC @waqarahmadshah
Muchas graciaass por tu comentario, revisaré ese enlace que me has dejado. Estos días, formando parte de dos comunidades que están en SEC, verdaderamente ha sido muy laborioso, pero seguimos..🙂
Oh gosh! Now that is serious... 2 communities in SEC.
You got all my sympathy my friend!!
I know what you need right now, a cup of coffee or two ☕☕
Thank you for inviting me. I have read the article whose link you shared. It was quite interesting and informative.
I don't want to talk further about AI topic. It's enough for today. If the Steemit team isn't taking this as seriously as they should, then it's not my headache to inform everyone. I'm not referring to you, but today I had a heated argument with a user on Discord that gave me a headache.
I haven't seen any response from the Steemit team or any detectives on my post. This issue is more alarming than it seems because I have just read the news about the announcement of GPT4, which is far more advanced than GPT3. I don't know where we are heading. 🤷♂️
I think it's time for us to learn from machines and improve ourselves before the machines replace us. 🙁
CC: @jesusjacr, @irawandedy, @kouba01
You are right... time to move on... time will tell
Lovely entry my friend, you have really written well
I wish you success 👍
Thank you
Your post has been successfully curated by @kouba01 at 45%.