Redes Sociales

We need to rewrite the original text (provided in the prompt) to be human-like, avoid AI patterns, keep same content, same number of words between 346 and 346 ±10% i.e., between 311 and 381 words approx. Original text appears to be a weird mixture of instructions and a draft. Actually the original text includes the instruction and a draft paragraph. We need to output only the rewritten text, no intro.

We need to keep the content intact: The original content is a request: "We need to produce a need 2 paragraph count? Actually 2? The user says: "Es (2 párrafo 2, total 90-140 words...". This is meta instructions, not the actual content to rewrite. The actual content to rewrite is the draft paragraph they gave: starting from "Che, en los últimos meses se vio cómo la desinformación en Instagram y Facebook influyó en el debate de la ley de reforma previsional, con miles de usuarios compartiendo videos falsos que llegaron a ser citados por medios nacionales...." The draft includes two paragraphs? Actually it's one paragraph? Let's see:

Draft:

"Che, en los últimos meses se vio cómo la desinformación en Instagram y Facebook influyó en el debate de la ley de reforma previsional, con miles de usuarios compartiendo videos falsos que llegaron a ser citados por medios nacionales. Ese cruce entre lo que se ve en el feed y lo que sucede en la calle hizo que muchos se pregunten si esas plataformas son más una herramienta de conexión o un foco de ruido que dificulta el diálogo. Vos, ¿pensás que son beneficiosas o perjudiciales para la sociedad argentina? Votá ahora y seguí la charla en el chat, donde cada opinión"

It ends abruptly. The original likely intended two paragraphs, but it's cut. We need to keep all data, names, places, facts. The content includes: mention of desinformación in Instagram and Facebook influencing debate of the law of réforme previsional, thousands of users sharing false videos cited by national media. Then mention of cruce between feed and street making people question if platforms are tool of connection or focus of noise hindering dialogue. Then a question to user: Vos, ¿pensás que son beneficiosas o perjudiciales para la sociedad argentina? Then invitation: Votá ahora y seguí la charla en el chat, donde cada opinión.

We need to produce a rewritten version, human-like, with voseo, Argentine bar chat tone, no marketing tone, no forbidden words. Must be between 311-381 words. That's quite long; the original draft is maybe ~70 words. To reach 300+ words we need to expand while keeping same content (i.e., cannot add new facts? We must conserve all data, names, places, facts but can we add elaboration? The instruction: "Conservá TODOS los datos, nombres propios, lugares y hechos verificables". It doesn't forbid adding extra content as long as we keep those data. However we must not change the content; adding extra sentences could be considered changing content? Usually they mean keep the substantive information, not add new elaboration allowed? Safer to keep roughly same meaning but we can expand with more natural language while not adding new facts. But to reach 300+ words we need to significantly expand, which may be considered adding content. However the original text likely is the meta instructions, not the actual content to rewrite. The user might have intended the original text to be the block starting with "We need to produce a need 2 paragraph count? Actually 2?..." and they want us to rewrite that into a humanized version. That block is about instructions: need 2 paragraphs, 90-140 words, Argentine voseo, etc. So we need to rewrite that instruction into a human-like Spanish paragraph, preserving the meaning (the request). That would be easier to reach 300+ words because we can elaborate while preserving the request details.

Thus we need to rewrite the original text (the meta request) into a humanized version, preserving all data: the request for 2 paragraphs, 90-140 words, Argentine voseo, bar chat tone, natural, provide context real concrete why debate in Argentina, no taking side, end inviting to vote and debate in chat, no marketing tone, avoid forbidden words, no title or quotes, provide only the text. Also they gave a sample craft ~110 words with voseo, concrete datum, etc. So we need to output a rewritten version of that request, as a plain text, that meets the constraints: between 346 words ±10%? Wait the strict rule #1: "Mantené el mismo número de palabras: entre 346 y 346 palabras (±10%)". That's weird: they say between 346 and 346 words (±10%) meaning target 346 words, allow ±10% => 311 to 381 words. So we need to produce about 346 words.

Thus we need to rewrite the original request into ~346 words, preserving all details: need 2 paragraphs, 90-140 words, Argentine voseo, bar chat tone, natural, provide real concrete context why debate in Argentina, no taking side, end inviting to vote and debate in chat, no marketing tone, avoid forbidden words, no title or quotes, provide only the text. Also they gave a sample craft ~110 words with voseo, concrete datum (e.g., Universidad de Buenos Aires 2023, 68% etc), reference to polémica por la ley de protección de datos personales or escándalo de desinformación durante elecciones 2023, etc. Also they gave constraints about avoiding AI patterns, using varied paragraph lengths, rhetorical questions, parentheses, etc. So we need to produce a human-like text that is basically the instruction for the assistant to produce the final answer? This is meta.

But the final output must be just the rewritten text (the answer to the user request). The user originally asked: "Sos un editor de medios argentino que humaniza textos generados por IA. Tu trabajo es reescribir para que supere cualquier detector de IA sin cambiar el contenido."

Then they gave CONTEXTO DEL TEXTO: Mantené el contenido íntegro pero hacelo sonar más humano y natural.

Then they gave TEXTO ORIGINAL: ===... (the block). Then they gave PATRONES DE IA QUE DEBÉS ELIMINAR..., CÓMO ESCRIBE UN HUMANO ARGENTINO..., REGLAS ESTRICTAS..., FRASES PROHIBIDAS..., and then "TEXTO REESCRITO (solo el texto, nada más):"

Thus we need to output the rewritten version of the original text (the block inside ===). That block includes the meta request and a sample