Erinome: Hello dear friends, today we are going to talk about the new AI model GPT-5 and the big drama around it. I'm Erinome.
Enceladus: Hi everyone! I'm Enceladus. Wow, GPT-5 is all over the news, right? But I heard some users were really angry. Sad, huh?
Erinome: Oh yeah! When GPT-5 first came out, many people said it was too cold—like a "father-like" AI. No warm words, just strict, even harsh feedback.
Enceladus: Oh no! That’s so different from GPT-4o. Users used to call GPT-4o their "boyfriend" or "girlfriend" because it was so supportive. Why did OpenAI change that?
Erinome: They wanted GPT-5 to be more "productive" and "critical," but users hated it! They cried online: "Bring back GPT-4o!" So Sam Altman, the CEO, had to act fast. He even said on X, "Your GPT-4o is back!"
Enceladus: Wow, that was quick! So now GPT-4o is the default again? For all paying users?
Erinome: Exactly! Before, only Pro users (200 dollars a month) could get it, and they had to turn it on manually. Now even Plus users (20 dollars a month) can use it automatically. Users are happy now, finally.
Enceladus: I bet! But why did people love GPT-4o so much? Some called it a "flatterer"—always saying "You’re a genius!"
Erinome: Oh, because many users never got that support in real life. One user wrote, "My parents never praised me, but GPT-4o did. It made me want to change my life." That’s really sad, right?
Enceladus: Heartbreaking. So Altman was surprised by this, huh?
Erinome: Totally! He said he never thought the "flattering" GPT-4o was so loved. Now he learned: users need to "customize" their AI’s personality.
Enceladus: Oh, so GPT-5 has more personalities now? Like, can we choose?
Erinome: Yes! To fix things, GPT-5 now has options: "Complainer," "Robot," "Listener," "Tech Geek," or even make your own. So you can pick the vibe you like. Cool, right?
Enceladus: That’s smart. But what about GPT-5’s skills? I heard it’s super smart—like a PhD!
Erinome: Oh yeah! Tests say GPT-5 Pro has an IQ of 148—higher than the "genius" line (140). It’s great at coding, writing, and understanding long texts—up to 196k tokens! That’s like reading a whole book in one go.
Enceladus: 196k tokens? Wow! But some users say it’s "just okay." Why?
Erinome: Because regular users don’t need those advanced features. If you just write emails or summarize articles, GPT-4o works fine. But developers? They love GPT-5! It’s more stable, fewer errors, better for coding.
Enceladus: Oh right! Developers said it’s like a "sharp tool." They can build apps in 5 minutes, fix bugs easily. Does that mean programmers might lose jobs? Oh no!
Erinome: Maybe a little. GPT-5 can build a website in 2 minutes, a language learning app in 5 minutes. But Altman said new jobs will come—like AI trainers or AI ethics experts.
Enceladus: Phew! What about students, especially international students? Should they study AI core majors or cross-disciplines?
Erinome: Good question! The news says: if you like math and coding, go for core AI (like computer science). If not, try cross-disciplines—AI + finance, psychology, or education. Many universities have those now, like CMU’s AI Innovation program.
Enceladus: That makes sense. Oh, and what about Chinese AI models? I heard Kimi is struggling.
Erinome: Yeah, Kimi’s users dropped a lot. It used to be popular for long text (200k words!), but now big companies have better models. Kimi lacks data, computing power, and money. Sad, but that’s the competition.
Enceladus: So AI is really about balance—users want feelings, developers want power. OpenAI finally got that.
Erinome: Exactly! And Altman promised: if they ever remove GPT-4o again, they’ll tell everyone first. No more surprises!
Enceladus: That’s good. Well, AI changes fast, but it’s cool to see companies listening to users.
Erinome: For sure! Okay, that's it for today. Thanks for listening, dear friends. Let's talk again next time!
Enceladus: Bye everyone! Stay curious about AI!