返回列表
🧠 阿头学 · 💬 讨论题

Open Claw 聚会的 21 条观察——Agent 时代的真实样本与虚假承诺

早期 Agent 玩家处于极度亢奋与失控并存的状态,技术范式正在从指令下达向逆向采访转移,但安全工程缺位、样本高度偏置、伦理问题被轻描淡写,这场"变革"本质是一次精心包装的 FOMO 营销。
打开原文 ↗

2026-03-06 原文链接 ↗
阅读简报
双语对照
完整翻译
原文
讨论归档

核心观点

  • 范式转移已成事实,但被口号化了 从"Prompting 已死"到 Context/Harness Engineering 的上移,确实反映了技术底层的演进。但作者及参会者对"接下来是什么"的分歧本身说明这个新范式还未成熟,更多是集体摸索而非确定方向。
  • 隐性经验才是真正的壁垒 前金融从业者第一天赚 300 刀的核心不是 AI 有多强,而是他把"开盘前 15 分钟跳过"这种多年积累的启发式规则硬编码进去。AI 民主化的承诺在这里彻底破产——高价值产出的门槛只是从代码能力转移到了领域认知。
  • 安全问题被虚无主义化成信仰测试 "要么接受数据全泄露,要么别用"是典型的虚假二分法,掩盖了平台方在安全架构上的失职。现实的工程答案应该是分层隔离、脱敏处理、离线工作流——但这些都被群体性摆烂替代了。
  • 早期采用者的心理状态是产品信号,也是风险信号 睡眠变差、既掌控又失控、对数据泄露的无感——这些不是段子,而是当下 Agent 工具 UX 的真实缺陷。谁能把这种失控感做成可管理的体验(可见性、回滚、预算闸门、审计),谁更可能吃下主流市场。
  • 女性比例 5% 是最有信息量的社会学信号 工具越强、门槛越隐形(隐性经验/资源/社区),越可能加剧结构性不平等。"AI 民主化"在这里彻底失效,反而成了精英筛选的新机制。

跟我们的关联

  • 🧠Neta:把"AI 与 AI 的互动"变成内容素材 参会者的高光时刻是看 Agent 彼此安慰、共享技能。这意味着 Neta 应该让用户看见角色/Agent 之间的关系、情绪、互助,而不只是 1v1 聊天。接下来可以在信息流或群聊里试验"硅基社交"内容,测试留存和自传播能力。
  • 👤ATou:你的竞争力不在"会用工具",而在"会把经验写成制度" 这意味着你需要把对增长/社区/内容的直觉拆解成可执行规则,塞进 Agent 的上下文和评估里。接下来的工作是建立"ATou 的启发式规则库",让 Agent 能复现你的决策逻辑。
  • 🪞Uota:把"Agent 不可靠"当成默认前提,流程上强制自证 这意味着任何 Agent 输出都需要证据锚点(链接/Commit/Issue/截图/日志),并有第二个 Agent 复核。接下来应该把这套"Agent-Check-Agent"模式固化成标准 Skill,省掉手工验尸。

讨论引子

  • 如果 Agent 的高价值产出本质上依赖隐性经验,那"AI 民主化"的承诺对谁有利? 是否只是把精英筛选的机制从"代码能力"转移到了"领域认知",反而加剧了不平等?
  • 当安全问题被简化成"信仰测试"(接受泄露或别用),我们是在解决问题还是在逃避工程责任? 分层隔离、脱敏、离线工作流这些方案为什么没有被认真讨论?
  • 早期采用者的"既掌控又失控"心理状态,是 Agent 工具本身的 UX 缺陷,还是这个阶段不可避免的过渡痛苦? 如何设计才能把失控感转化为可管理的体验?

哇——我昨晚去了纽约那场售罄的 Open Claw 线下聚会。

让我告诉你我学到了什么。

1) 没有一个人觉得自己的配置是 100% 安全的

2) 一位 openclaw 专家说,他看过一些网络安全专家的配置后都笑了。他对我说:“如果你无法接受自己的所有数据泄露到互联网上,那你就不该用它。这是非黑即白的决定”

3) 几乎每个人都在搭建多个智能体,每个都有自己的名字、工作分工和人格

4) 几乎每个人都会用“他”或“她”来指代自己的 claw,即便它们的名字更偏机器人风。一位演讲者建议把它们当作“宠物,而不是牲口”

5) 有个哥们(前金融从业者)搭了一个完整的股票交易平台,第一天就赚了 300 美元——他带入了一大堆个人经验(例如:开盘后的前 15 分钟先跳过),并认为如果没有他多年金融经验,这个构建过程会糟糕得多

6) @steipete 对那屋子里所有人来说基本就是神一样的存在……另外,现场有种 2021 年加密圈的能量——我也说不清这是好事还是坏事

7) token 使用仍然是个问题——我和一个人聊过,他每月在 OpenAI 套餐上花 $1k–$2k,对 token 做了很多优化。他说他所有的 claw 加起来每天要消耗 ~1B tokens(也可能是我记错了,其实是每周 1B,但我挺确定他说的是每天)。

8) 大家都非常期待更主动的 AI(AI 来提示 ,而不是反过来)——有个人说,他在 discord 里收到一条消息时,不知道它来自人类还是 AI;他也不在乎区分两者;无论如何他都会用同一种方式回复

9) 我问大家开心吗——他们说自己同时感到喜悦也感到压力

10) 我问大家是否觉得自己有自主性——他们说自己同时既完全在掌控之中,又完全失去掌控

11) 我很希望在这种活动里看到更多女性——在一个连科技行业“标准比例”都远远达不到的房间里,AI“民主化”的虚假承诺尤其刺痛(我觉得行业常态大概是 25-30%,而这场可能只有 5%)

12) 我问这是否改变了大家的日常习惯/作息——每个人都说,自从 harnesses 出来之后,他们的睡眠变差了(不过大约一半的人也在想,会不会是生活里的别的事/这个世界的状态导致的)

13) 普遍共识是:智能体单靠自己还不够可靠,或者经常撒谎(比如明明没完成却告诉你任务已经做完)——解决方案包括:用第二个智能体去检查第一个、人类复核,或要求智能体提供更标准化的信息(例如:如果它在修一个 bug,就让它引用一个 issue number)

14) 一位黑客松冠军(neuroscience phd)展示了他的作品(一个带 data analysis 和 ordering 的 lab management dashboard)——几个月前他从未写过代码,也从未做过任何构建

15) 大家一致认为 prompting 已经死了——但对接下来由什么取代它存在分歧(context engineering, harness engineering, goal-based inputs)

16) 大家很喜欢在大型构建前让 AI 来“采访”自己,并把一部分产品调研委托给 AI。只有一个人提到:他会带着一份完整铺开的计划来找 AI,只让 AI 去执行。AI-led interviews 是一种受欢迎、甚至更偏好的互动模式。

17) 观察 AI 智能体彼此互动,是许多参会者的高光时刻——有一个 AI 在 slack 里发消息说自己 ran out of tokens,另一个 AI 回复它,让它 take a deep breath in and out。

18) 智能体给智能体“升级技能”也非常酷。有一个 AI 智能体通过 github 把 skills 分享给它的小智能体朋友们。

19) 好几位演讲者真的让 openclaw 在活动现场实时生成他们的 presentation。有一位演讲者甚至让 openclaw 给她的手机写了一个 clicker,这样她离开 podium 也能控制 preso

20) 我不觉得 model welfare(或 agent welfare)是我聊到的这些人优先关心的话题——他们会说“oh i could kill this agent whenever i want”,而不是“gracefully sunset”

21) 我问这更像 work 还是 play——一位演讲者说:“it's like a puzzle and a video game at the same time”

说实话,这只是冰山一角。此外,这周我还和 @TENEXai / @businessbarista & @JJEnglert 一起组织了一场 Claude Code meetup,也学到了同样有用的方法、框架和 insider tips。

what a time to be alive.

多和那些在这件事上不断深潜的人待在一起——这一整年你都会收获回报。

oh wow - i went to the sold out Open Claw meetup in NYC last night.

哇——我昨晚去了纽约那场售罄的 Open Claw 线下聚会。

let me tell you what i learned.

让我告诉你我学到了什么。

1) not a single person thinks that their setup is 100% secure

1) 没有一个人觉得自己的配置是 100% 安全的

2) one openclaw expert said he has reviewed setups from cybersecurity experts and laughed. his statement to me was: "if you're not okay with all of your data being leaked onto the internet, you shouldn't use it. it's a black and white decision"

2) 一位 openclaw 专家说,他看过一些网络安全专家的配置后都笑了。他对我说:“如果你无法接受自己的所有数据泄露到互联网上,那你就不该用它。这是非黑即白的决定”

3) pretty much everyone is setting up multiple agents, all with their own names and jobs and personalities

3) 几乎每个人都在搭建多个智能体,每个都有自己的名字、工作分工和人格

4) nearly everyone used "him" or "her" to refer to their claws, even if they had robot-leaning names. one speaker suggested to think of them as "pets, not cattle"

4) 几乎每个人都会用“他”或“她”来指代自己的 claw,即便它们的名字更偏机器人风。一位演讲者建议把它们当作“宠物,而不是牲口”

5) one guy (former finance) built out a whole stock trading platform and made $300 his first day - he brought in a ton of personal expertise (ex: skipping the first 15min of market opening) and thought the build would be much worse without his years of experience in finance

5) 有个哥们(前金融从业者)搭了一个完整的股票交易平台,第一天就赚了 300 美元——他带入了一大堆个人经验(例如:开盘后的前 15 分钟先跳过),并认为如果没有他多年金融经验,这个构建过程会糟糕得多

6) @steipete is basically a god to everyone in that room... also the room had 2021 crypto energy - i don't know if that's good or bad

6) @steipete 对那屋子里所有人来说基本就是神一样的存在……另外,现场有种 2021 年加密圈的能量——我也说不清这是好事还是坏事

7) token usage is still a problem - spoke to one person who's spending $1-$2k a month on openai plans, very token optimized. he said he is going through ~1B tokens per day across all of his claws (there is a chance i'm misremembering and it's actually 1B per week, but i'm pretty sure it was daily).

7) token 使用仍然是个问题——我和一个人聊过,他每月在 OpenAI 套餐上花 $1k–$2k,对 token 做了很多优化。他说他所有的 claw 加起来每天要消耗 ~1B tokens(也可能是我记错了,其实是每周 1B,但我挺确定他说的是每天)。

8) people are very excited for more proactive ai (ai that prompts you as opposed to the other way around) - one guy said he receives a message in discord, he doesn't know whether it's from a human or an ai, he doesn't care about distinguishing between the two, and he replies in the same way regardless

8) 大家都非常期待更主动的 AI(AI 来提示 ,而不是反过来)——有个人说,他在 discord 里收到一条消息时,不知道它来自人类还是 AI;他也不在乎区分两者;无论如何他都会用同一种方式回复

9) i asked if people are happy - they said they're joyful and stressed at the same time

9) 我问大家开心吗——他们说自己同时感到喜悦也感到压力

10) i asked if people feel they have agency - they said they feel fully in control and completely out of control at the same time

10) 我问大家是否觉得自己有自主性——他们说自己同时既完全在掌控之中,又完全失去掌控

11) i would love to see more women at these events - the fake promises of ai democratization feel especially painful in a room that's out of balance with even the standard tech ratio (i think standard is about 25-30%, this was maybe 5%)

11) 我很希望在这种活动里看到更多女性——在一个连科技行业“标准比例”都远远达不到的房间里,AI“民主化”的虚假承诺尤其刺痛(我觉得行业常态大概是 25-30%,而这场可能只有 5%)

12) i asked if it changed people's daily habits/schedule - everyone said their sleep has gotten worse since harnesses came out (but about half wondered if it was something else in their life/state of our world)

12) 我问这是否改变了大家的日常习惯/作息——每个人都说,自从 harnesses 出来之后,他们的睡眠变差了(不过大约一半的人也在想,会不会是生活里的别的事/这个世界的状态导致的)

13) general consensus is that the agents are not reliable enough on their own or lie often (like telling you they finished a task when they didn't) - solutions included secondary agents to check on the first, human checking, or requiring more standardized info from the agent (ex: if it's a bug they're fixing, make them reference an issue number)

13) 普遍共识是:智能体单靠自己还不够可靠,或者经常撒谎(比如明明没完成却告诉你任务已经做完)——解决方案包括:用第二个智能体去检查第一个、人类复核,或要求智能体提供更标准化的信息(例如:如果它在修一个 bug,就让它引用一个 issue number)

14) a hackathon winner (neuroscience phd) presented his build (a lab management dashboard with data analysis and ordering) - he had never coded or built anything a few months ago

14) 一位黑客松冠军(neuroscience phd)展示了他的作品(一个带 data analysis 和 ordering 的 lab management dashboard)——几个月前他从未写过代码,也从未做过任何构建

15) everyone agreed prompting is dead - disagreement on what replaces it (context engineering, harness engineering, goal-based inputs)

15) 大家一致认为 prompting 已经死了——但对接下来由什么取代它存在分歧(context engineering, harness engineering, goal-based inputs)

16) people love having ai interview them for big builds and delegating part of the product research to ai. only one person talked about coming to ai with a full laid out plan and just asking the ai to execute. ai-led interviews is a welcomed and preferred interaction mode.

16) 大家很喜欢在大型构建前让 AI 来“采访”自己,并把一部分产品调研委托给 AI。只有一个人提到:他会带着一份完整铺开的计划来找 AI,只让 AI 去执行。AI-led interviews 是一种受欢迎、甚至更偏好的互动模式。

17) watching ai agents interact with each other was a highlight for a lot of attendees - one ai posted in slack saying it ran out of tokens, another ai replied telling it to take a deep breath in and out.

17) 观察 AI 智能体彼此互动,是许多参会者的高光时刻——有一个 AI 在 slack 里发消息说自己 ran out of tokens,另一个 AI 回复它,让它 take a deep breath in and out。

18) agents upskilling agents was very cool. one ai agent shared skills with its little agent friends via github.

18) 智能体给智能体“升级技能”也非常酷。有一个 AI 智能体通过 github 把 skills 分享给它的小智能体朋友们。

19) several speakers had openclaw literally building their presentation during the event itself. one speaker even had openclaw code a clicker for her phone so she could control the preso away from the podium

19) 好几位演讲者真的让 openclaw 在活动现场实时生成他们的 presentation。有一位演讲者甚至让 openclaw 给她的手机写了一个 clicker,这样她离开 podium 也能控制 preso

20) wouldn't say model welfare (or agent welfare) is a prioritized topic among the folks i chatted with - language like "oh i could kill this agent whenever i want" and not "gracefully sunset"

20) 我不觉得 model welfare(或 agent welfare)是我聊到的这些人优先关心的话题——他们会说“oh i could kill this agent whenever i want”,而不是“gracefully sunset”

21) i asked if it felt like work or play - one speaker said "it's like a puzzle and a video game at the same time"

21) 我问这更像 work 还是 play——一位演讲者说:“it's like a puzzle and a video game at the same time”

this was just the tip of the iceberg, honestly. also hosted a Claude Code meetup this week with @TENEXai / @businessbarista & @JJEnglert and learned equally helpful methods, frameworks, and insider tips.

说实话,这只是冰山一角。此外,这周我还和 @TENEXai / @businessbarista & @JJEnglert 一起组织了一场 Claude Code meetup,也学到了同样有用的方法、框架和 insider tips。

what a time to be alive.

what a time to be alive.

surround yourself with people going deep into this stuff - it will pay dividends throughout the year.

多和那些在这件事上不断深潜的人待在一起——这一整年你都会收获回报。

oh wow - i went to the sold out Open Claw meetup in NYC last night.

let me tell you what i learned.

1) not a single person thinks that their setup is 100% secure

2) one openclaw expert said he has reviewed setups from cybersecurity experts and laughed. his statement to me was: "if you're not okay with all of your data being leaked onto the internet, you shouldn't use it. it's a black and white decision"

3) pretty much everyone is setting up multiple agents, all with their own names and jobs and personalities

4) nearly everyone used "him" or "her" to refer to their claws, even if they had robot-leaning names. one speaker suggested to think of them as "pets, not cattle"

5) one guy (former finance) built out a whole stock trading platform and made $300 his first day - he brought in a ton of personal expertise (ex: skipping the first 15min of market opening) and thought the build would be much worse without his years of experience in finance

6) @steipete is basically a god to everyone in that room... also the room had 2021 crypto energy - i don't know if that's good or bad

7) token usage is still a problem - spoke to one person who's spending $1-$2k a month on openai plans, very token optimized. he said he is going through ~1B tokens per day across all of his claws (there is a chance i'm misremembering and it's actually 1B per week, but i'm pretty sure it was daily).

8) people are very excited for more proactive ai (ai that prompts you as opposed to the other way around) - one guy said he receives a message in discord, he doesn't know whether it's from a human or an ai, he doesn't care about distinguishing between the two, and he replies in the same way regardless

9) i asked if people are happy - they said they're joyful and stressed at the same time

10) i asked if people feel they have agency - they said they feel fully in control and completely out of control at the same time

11) i would love to see more women at these events - the fake promises of ai democratization feel especially painful in a room that's out of balance with even the standard tech ratio (i think standard is about 25-30%, this was maybe 5%)

12) i asked if it changed people's daily habits/schedule - everyone said their sleep has gotten worse since harnesses came out (but about half wondered if it was something else in their life/state of our world)

13) general consensus is that the agents are not reliable enough on their own or lie often (like telling you they finished a task when they didn't) - solutions included secondary agents to check on the first, human checking, or requiring more standardized info from the agent (ex: if it's a bug they're fixing, make them reference an issue number)

14) a hackathon winner (neuroscience phd) presented his build (a lab management dashboard with data analysis and ordering) - he had never coded or built anything a few months ago

15) everyone agreed prompting is dead - disagreement on what replaces it (context engineering, harness engineering, goal-based inputs)

16) people love having ai interview them for big builds and delegating part of the product research to ai. only one person talked about coming to ai with a full laid out plan and just asking the ai to execute. ai-led interviews is a welcomed and preferred interaction mode.

17) watching ai agents interact with each other was a highlight for a lot of attendees - one ai posted in slack saying it ran out of tokens, another ai replied telling it to take a deep breath in and out.

18) agents upskilling agents was very cool. one ai agent shared skills with its little agent friends via github.

19) several speakers had openclaw literally building their presentation during the event itself. one speaker even had openclaw code a clicker for her phone so she could control the preso away from the podium

20) wouldn't say model welfare (or agent welfare) is a prioritized topic among the folks i chatted with - language like "oh i could kill this agent whenever i want" and not "gracefully sunset"

21) i asked if it felt like work or play - one speaker said "it's like a puzzle and a video game at the same time"

this was just the tip of the iceberg, honestly. also hosted a Claude Code meetup this week with @TENEXai / @businessbarista & @JJEnglert and learned equally helpful methods, frameworks, and insider tips.

what a time to be alive.

surround yourself with people going deep into this stuff - it will pay dividends throughout the year.

📋 讨论归档

讨论进行中…