Look like ChatGPT | freedom

Join any Zoom call, walk in any lecture hall, watch any video on YouTube, and listen carefully. After the content and inside the linguistic patterns, you will find creeping monotheism of the sound of artificial intelligence. Words like “ingenuity” and “fabric”, preferred by Chatgpt, sneaks to our vocabulary, while words such as “reinforcement”, “detection”, “nuance”, the words that Chatgpt preferred were less used. Researchers are already documenting the transformations in the way we talk and continue as a result of a chat – and they see this linguistic effect accelerating something much larger.

Within 18 months after Chatgpt, speakers used words such as “accuracy”, “Defer”, “Realm” and “ADEPT” up to 51 percent more than it was in the previous three years, according to researchers at the Max Planck Institute for Human Development, Who analyzed nearly 280,000 video clips on YouTube From academic channels. The researchers excluded other possible change points before the ChatGPT release and confirmed that these words are in line with those preferred by the model, as shown in A previous study compares 10,000 human and specific texts. The speakers do not realize that their language changes. This is exactly the point.

One word, in particular, has emerged for researchers as a kind of linguistic watermark. “Defer” Shibboleth, the Neon brand in the middle of every conversation Chatgpt was here. “We absorb these virtual vocabulary in daily communication,” says Herumo Yakura, the main author of the post -PhD study and researcher at the Max Planck Institute for Human Development.

But not only that we adopt the language of artificial intelligence – it is about how to start the sound. Although current studies are mostly focused on vocabulary, researchers doubt that the effect of artificial intelligence is also appearing in a tone – in the form of a longer and more organized speech and silent emotional expression. Livin Bernckman, a research scientist at the Max Planck Institute for Human Development and a co -author of the study, also places “Defer” is just a tip of the iceberg.

Artificial intelligence appears clearly in functions such as smart responses, automatic correction, and spelling. Search from Cornell We look at our use of smart responses in the chats, finding that the use of smart responses increases comprehensive cooperation and feelings of rapprochement between the participants, because users end up choosing a more positive emotional language. But if people believed that their partner was using artificial intelligence in the interaction, they classified their partner as less cooperative and more demanding. It was not in place, it was not the use of actual artificial intelligence that stopped them – the doubt was. We constitute perceptions based on language signals, which are in fact the characteristics of the language that pays these impressions.

This paradox – Amnesty International refers to improving communication while promoting suspicion – to losing deeper confidence, according to Maur Naman, professor of information science at Cornell Tech. It has identified three levels of human signs that we lost in adopting artificial intelligence in our connection. The first level is that the basic human signs, and the sermon that speaks to our authenticity as a human being like moments of weakness or personal rituals, which say to others, “This is me, I am a human being.” The second level consists of the attention and effort that proves, “I was interested in enough to write this myself.” The third level is the ability signs that show our sense of humor, our competence, and the true ourselves of others. This is the difference between sending a text message to someone, “I am sorry because you are upset” for “Sorry I felt horror at dinner, and I may not have skipped treatment this week.” One looks flat for other human sounds.

For Naaman, knowing how to repeat these signs and raise them is the way forward in communication by artificial intelligence, because artificial intelligence is not only the change of language-but what we think about. “Even in dating sites, what does it mean to be funny in your profile or in chatting anymore where we know that artificial intelligence can be funny to you?” Naman asks. The agency’s loss from our discourse and the transition to our thinking, in particular, is what worries it. “Instead of clarifying our own ideas, we express anything that helps us to express … We have become more convincing.” Without these signals, Naaman warns, we will only trust in contact with face-not even video calls.

The problem of confidence is when you think that artificial intelligence quietly proves who gets a “legal” voice in the first place. University of California, Berkeley Research I found that artificial intelligence responses often contain stereotypes or inaccurate approximation when demanding the use of accents other than the standard American English. Examples of this include Chatgpt repeatedly provided to the non-American non-American user-because of the lack of understanding and exaggeration of the entry tone. One of the English respondents, Singaporean Stuck“Singlish was a little bit worthy of responses.” The study revealed that artificial intelligence does not only prefer the standard American English, but rather is actively falling in other dialects in ways that can block loudspeakers.

This system perpetuates inaccuracy not only about societies but also about what the English language is “correct”. So the risks are not only related to preserving linguistic diversity – they are about to protect defects that already build confidence. When everyone around us begins to sound “correct”, we lose verbal obstacles, regional expressions, and phrases outside the crackers that indicate weakness, originality and personality.

We are approaching the division point, as the effects of artificial intelligence affect how to speak and write the move between the pillars of the standard, such as the arrangement of vocational emails or official presentations, and the original expression in personal and emotional spaces. Among those columns, there are three basic tensions in play. Early reaction signals, such as the avoid “Defer” academics and people who actively try to not appear like artificial intelligence, indicate that we may organize self -regulation against homogeneity. Artificial intelligence systems are likely to become more expressive and customized over time, which may reduce the current problem of artificial intelligence. The deepest danger to all, as Naman pointed out, is not linguistic monotheism, but it is the loss of conscious control over our thinking and expression.

The future has not been determined between homogeneity and hyper: this depends on whether we are aware of this change. We see early signs that people will retreat when the effect of artificial intelligence becomes very clear, while technology may develop to better reflect human diversity rather than flatten. This is not a question about whether Amnesty International will continue to form how to speak – because it will work – but whether we will choose actively to maintain the space for verbal justifications and emotional chaos that make communication identifying, unknown.