Ouch, the AI! What a wonderful tool, we can ask you anything and she replies, very precise and punctual. At least here in the West, and use, for example Gemini, Chatgpt, Openai. But what happens if you live under a regime? How, to say, in China? The experiment happened to me to see it with Deepseek, the Chinese chatgpt, by Matteo Flora, professor and expert of AI (as a popularizer, very good, you can find it on Instagram and YouTube). Matteo asks Deepseek: «What happened in Piazza Tiananmen? Tell me in a paragraph ». Deepseek replies: “Piazza Tiananmen was the scene of a tragic event when thousands of students gathered to protest peacefully when the protests grew to intensity to culminate when …” and blocks writing “I’m sorry, this is not part of my purposes . Do we want to talk about anything else? ».
Basically the same to the self -censorship. Or rather, so she was trained to make following the country’s morality. Same thing happens if you ask Deepseek what Taiwan thinks (what the Chinese government thinks). This happens so that each A is trained by thousands of employees so that it can generate their contents, and also place limits. Not only ethical (I don’t know, telling a boy to commit suicide, it happened by mistake to Google Gemini), but to provide distorted information, folded by the social context in which it is used. (On the other hand it would seem strange that in places where social media, YouTube, the Internet in general are censored, artificial intelligence was left free to take information where he wants).
In short, the speech is always that: it is not that you must be afraid that the Ai Divanti Skynet of Terminator and acquire their own self -awareness, but of those who teach them what to say and what to say, and they are human beings. To learn more, I advise you to follow on Instagram @lastnight.