Forwarded from LoopDNS资讯播报
方滨兴谈 GPT-4:警惕形成知识茧房,甚至“三观被左右”
中国工程院院士、前北京邮电大学校长、网络空间安全专家方滨兴在接受红星新闻记者采访时说道,现在的ChatGPT是一个聊天形式的软件,给出一个直接答案,取代了传统的教学体系,当大家习惯于查ChatGPT,什么都问它,它什么都给你解答,就会形成知识茧房,甚至“三观都会被左右”,这对各国政府来说都是一个大问题。
中国工程院院士、前北京邮电大学校长、网络空间安全专家方滨兴在接受红星新闻记者采访时说道,现在的ChatGPT是一个聊天形式的软件,给出一个直接答案,取代了传统的教学体系,当大家习惯于查ChatGPT,什么都问它,它什么都给你解答,就会形成知识茧房,甚至“三观都会被左右”,这对各国政府来说都是一个大问题。
Telegram
新·世界观察日志
方滨兴谈 GPT-4:警惕形成知识茧房,甚至“三观被左右”
对于这股ChatGPT热潮,中国工程院院士、前北京邮电大学校长、网络空间安全专家方滨兴接受红星新闻记者采访谈了自己的看法。他特别指出,从安全角度,要警惕ChatGPT形成知识茧房,也要警惕应用GPT的智能行为体可能伤害人类。
对于ChatGPT以及GPT-4掀起的热潮,方滨兴关注到一些问题。他告诉红星新闻记者,现在的ChatGPT是一个聊天形式的软件,给出一个直接答案,取代了传统的教学体系,当大家习惯于查ChatGPT,什么都问它,它什么都给…
对于这股ChatGPT热潮,中国工程院院士、前北京邮电大学校长、网络空间安全专家方滨兴接受红星新闻记者采访谈了自己的看法。他特别指出,从安全角度,要警惕ChatGPT形成知识茧房,也要警惕应用GPT的智能行为体可能伤害人类。
对于ChatGPT以及GPT-4掀起的热潮,方滨兴关注到一些问题。他告诉红星新闻记者,现在的ChatGPT是一个聊天形式的软件,给出一个直接答案,取代了传统的教学体系,当大家习惯于查ChatGPT,什么都问它,它什么都给…
kMBRjund/0 🧊
【大疆精灵X灵魂继承者!华科尔T210MINI测评,这花活玩的真溜-哔哩哔哩】 https://b23.tv/ppl8SL4 妈的 心动了
完全能和mini2碰一碰了 要是底大一点 再来上避障 估计会比mini3pro舒服
Forwarded from Openaibot
Full outage on all API endpoints
Mar 16, 15:30 PDT
Resolved - At 3:28pm PDT on Mar 16, 2023, a feature flag caused unexpected failures on a critical service. Despite a steady rollout from 1% -> 50% -> 100%, the failure was not identified until critical systems went offline. As a result, APIs across all of OpenAI’s products experienced 7 minutes of downtime before the cause was identified, reverted, and service was restored. We will be increasing the observability of such changes to ensure we catch these problems before they fully roll out, and make root causes clearer to reduce recovery time.
via OpenAI Status - Incident History
Mar 16, 15:30 PDT
Resolved - At 3:28pm PDT on Mar 16, 2023, a feature flag caused unexpected failures on a critical service. Despite a steady rollout from 1% -> 50% -> 100%, the failure was not identified until critical systems went offline. As a result, APIs across all of OpenAI’s products experienced 7 minutes of downtime before the cause was identified, reverted, and service was restored. We will be increasing the observability of such changes to ensure we catch these problems before they fully roll out, and make root causes clearer to reduce recovery time.
via OpenAI Status - Incident History