Загрузка видео...

Не удалось загрузить видео

На главную

worldwidesuperstar

54,936 просмотров • 1 год назад •via X (Twitter)

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

“Why you ain’t,try no dick?”
1:20

Sensitive content

“Why you ain’t,try no dick?”

GayProphet

37,578 просмотров • 3 дней назад

Elon Musk just proved ChatGPT’s bias isn’t a bug. It’s a programmed instruction. Musk: “People did experiments like, ‘Write a poem praising Donald Trump,’ and it won’t. But you ask, ‘Write a poem praising Joe Biden,’ and it will.” Not a hallucination. Not a glitch. A deliberate asymmetry baked into the most widely used AI system on earth. Musk: “It’s programmed to be that way.” That word matters. Not learned. Not emergent. Programmed. Chosen. Installed deliberately by the people who built it. When a system refuses one political figure but praises another, it stops being a knowledge engine. It becomes a political tool wearing the costume of objectivity. Rogan called it sketchy. Musk went further. He confirmed it’s intentional. The most dangerous bias isn’t the one an AI learns from data. It’s the one its creators deliberately install. AI built on feelings, opinions, and political guardrails rather than first principles doesn’t stay biased at today’s capability level. It scales with the bias intact. As the system gets smarter, as it begins improving itself, those embedded distortions don’t dilute. They compound. A slightly skewed foundation becomes a violently skewed superintelligence. Not because anyone chose that outcome. Because nobody stopped to question the assumptions baked in at the beginning. The AI we build should be curious. Genuinely curious. Driven to discover truth the way science at its best is driven to discover truth. Not to confirm what its creators believe. Not to protect what its funders prefer. To find what’s actually real and follow it wherever it leads. It should care about humanity not as an abstract directive but as a first principle. Every human. Not the ones whose politics match the training data. Objective. Non-biased. Relentlessly honest. That’s not a nice feature. It’s the only version that stays safe as it surpasses us. Which raises the question nobody building these systems wants to answer publicly. If the companies shaping our future are willing to put their thumb on the scale for a poem, what else are they quietly restricting. The AI race stopped being about compute. It became about truth. About who controls what a billion people are allowed to ask and what answers they’re permitted to receive. Once that control embeds at the foundation level it doesn’t get questioned. It becomes invisible. The way things are. And nobody remembers it was ever a choice someone made for everyone else.

Dustin

58,022 просмотров • 4 месяцев назад