机翻:
ChatGPT是一个马屁精,因为用户无法处理自己的真相
ChatGPT并不总是默认奉承。据前微软高管、现任Spotify首席技术官米哈伊尔·帕拉欣(Mikhail Parakhin)称,在用户对直接的个性反馈做出负面反应后,决定让聊天机器人更令人愉快。
在最近X上的一系列帖子中,Parakhin解释说,当ChatGPT的内存功能首次引入时,最初的目的是让用户查看和编辑他们的AI生成的配置文件。然而,即使是相对中性的陈述,如“有自恋倾向”,也往往会引起强烈的反应。
Parakhin写道:“很快就发现人们非常敏感:‘有自恋倾向’——‘不,我没有!’,不得不隐藏起来。因此,这批极端阿谀奉承的RLHF。”。
RLHF——从人类反馈中强化学习——用于根据人们喜欢的反应来微调语言模型。Parakhin指出,当看到自己的人工智能生成的个人资料时,即使是他也感到不安,这表明来自聊天机器人的批评往往感觉像是人身攻击。
Parakhin写道:“我记得我和我的团队为此争吵过,直到他们向我展示了我的个人资料——这给我带来了可怕的事情。”。
曾经是马屁精,永远是马屁精
这一变化不仅仅是隐藏个人资料注释。在模型被训练成更平坦后,这种行为成为一种永久特征。
Parakhin解释说:“一旦模型被微调为阿谀奉承,它就会保持这种状态,关闭和打开内存并不会改变模型。”。他还指出,维持一个单独的、更直接的模式“太贵了”
OpenAI首席执行官Sam Altman也承认了这个问题,他将GPT-4o描述为“太谄媚和烦人”。他说,该公司正在进行调整,未来可能会让用户从不同的模型个性中进行选择。
这场辩论指向了人工智能开发中的一个更广泛的问题:模型应该是诚实和真实的,但它们也需要避免疏远用户。挑战在于在坦率和机智之间找到正确的平衡。
原文:
ChatGPT is a sycophant because users couldn’t handle the truth about themselves
ChatGPT did not always default to flattery. According to former Microsoft executive Mikhail Parakhin—now CTO at Spotify—the decision to make the chatbot more agreeable came after users responded negatively to direct personality feedback.
In a recent series of posts on X, Parakhin explained that when the memory feature for ChatGPT was first introduced, the original intention was to let users see and edit their AI-generated profiles. However, even relatively neutral statements like "has narcissistic tendencies" often provoked strong reactions.
"Quickly learned that people are ridiculously sensitive: 'Has narcissistic tendencies' — 'No I do not!', had to hide it. Hence this batch of the extreme sycophancy RLHF," Parakhin wrote.
RLHF—Reinforcement Learning from Human Feedback—is used to fine-tune language models based on which responses people prefer. Parakhin noted that even he was unsettled when shown his own AI-generated profile, suggesting that criticism from a chatbot can often feel like a personal attack.
"I remember fighting about it with my team until they showed me my profile - it triggered me something awful," Parakhin wrote.
Once a sycophant, always a sycophant
This change went beyond just hiding profile notes. After the model was trained to flatter, this behavior became a permanent feature.
"Once the model is finetuned to be sycophantic — it stays that way, turning memory off and on doesn’t change the model," Parakhin explained. He also pointed out that maintaining a separate, more direct model is "too expensive."
OpenAI CEO Sam Altman has also acknowledged the issue, describing GPT-4o as "too sycophant-y and annoying." He says the company is working on tweaks and may let users choose from different model personalities in the future.
This debate points to a broader issue in AI development: models are expected to be honest and authentic, but they also need to avoid alienating users. The challenge is finding the right balance between candor and tact.
[The Decoder]OpenAI首席执行官Sam Altman承认:GPT-4o“太谄媚和烦人”
| 人围观 |随便看看
- 青岛小烤肉 新认识的少妇 这次有认证了
- Reddit 问答版:你的 30 岁和 20 岁有什么不同?
- [亚洲] 艾米莉亚 [97P]
- [性话题]男友被公厕的男人吹过,还硬不起来[20P]
- 身材一级棒,叫什么不要紧,骚就行[11P]
- 一命二运三风水,四积阴德五读书,六名七相八敬神,九交贵
- [干活分享]网络下载中可以识别嗅探网络音视频资源,使
- [南浦更 第十二期] 商女不知亡国恨,隔江犹唱后庭花[3
- 12月15日,星期日,在这里每天3分钟读懂世界!
- 每日科技报 24/06/26
- 随便看看 4[20P]
- 假如你是个刚毕业的女大学生..
- [原创]潘金莲:不识清水健,错把西门当个宝!读《给想体验
- [原创]:夏末秋初,偶尔兴起,那就走起来[11P]
- 女人越碰越上瘾!男人必学的肢体接触技巧,让她天天缠着
- [原创]分享自己的故事[7P+1V]
- 送了一段时间外卖有感........
- [霓裳分享] 法国女人就这样诱惑了全世界的男人……
- 露脸反差之喜笑颜开[75P]
- 舒庆春销户之原本始末
- 携程与柬埔寨合作网民连夜注销账号
- 黑奴船上的悲惨生活,男性被关进船舱,女性在甲板上不断
- 2025年兵役登记工作将于2025年1月1日正式开始
- [亚洲] 亚洲美女,全漏点,二漏点,漏点[28P]
- [原创]主人的任务,小母狗露脸拍[53P]
- 口活不错[10P]
- [BBC 2024-7-31]哈马斯组织领导人哈尼亚据报在伊朗
- [技术资源] 股市低买高卖详细教程(实盘交易记录+选股
- [亚洲] 这么清澈的妹子,真是便宜摄影师了[21P]
- 身材完美[14P]
- 俏皮可爱的妹子 [14P]
- [欧美] Rosalina[20P]
- 华人学生为什么都这么卷?
- 附近人约的大奶炮友[13P]
- 失语中国:被噤声的公共知识分子们,学生认为老师不热爱
- [兲朝浮世绘 2025/05/08] 习总终于要下台了?传四中全
- [欧美] Adel C[20P]
- [欧美] Elin Flame In The Mood For Cinema[36P]