再见,Gmail
Bye Bye Gmail

原始链接: https://m24tom.com/bye-bye-gmail/show

## 邮箱地址变更及对人工智能训练的担忧 请将我的主要邮箱地址更新为已移除,替换我之前的@gmail.com地址。我会尽量查看旧账户,但无法保证。 此次变更源于我对谷歌将人工智能整合到Gmail中的日益不安。虽然我早期就采用了大型语言模型,但我开始担心谷歌使用我的邮件内容来训练模型——特别是机密信息泄露和真实沟通被侵蚀的风险。 我珍视阅读朋友们精心撰写的邮件,这是快速短信或现在无法禁用的AI摘要所无法比拟的。最近,Gmail开始提示我“分享”邮件给谷歌,明确表示这些邮件*可能*会被人工审核用于模型训练。 我正在转向微软的邮件托管服务,希望获得更多隐私,ProtonMail 也是一个潜在的未来选择。我感谢多年来使用Gmail的时光,但我不愿意贡献我的个人写作风格和数据来进一步发展人工智能。

## 再见Gmail:数据训练的转变 最近一篇Hacker News上的讨论,由一位用户(“tklenke”)放弃Gmail引发,凸显了人们对谷歌利用电子邮件内容训练其人工智能模型的担忧。这位作者是一位长期Gmail用户,对强制的人工智能摘要以及要求共享数据以供人工审查的提示感到不安——本质上,允许谷歌利用个人和机密信息进行LLM训练。 虽然作者承认谷歌过去曾将电子邮件用于定向广告,但他们认为训练人工智能模型是更进一步。他们已迁移至微软作为临时解决方案,ProtonMail则作为潜在的长期替代方案。 对话强调了拥有自己的域名以实现电子邮件独立的重要性,Fastmail、Migadu和Purelymail等服务经常被推荐。许多评论者分享了成功的迁移故事,强调了迁移过程出乎意料的简单以及控制权的益处。 也有一些用户指出禁用“智能功能”并使用独立电子邮件客户端以获得更好的隐私和组织管理的重要性。 最终,这场讨论反映了人们对个人数据控制权的渴望,以及对大型科技公司数据实践的“学会接受”心态的拒绝。
相关文章

原文

Friends,

TLDR: Please remove @gmail.com as my primary email address and substitute removed. I will try to respond to email sent to @gmail.com but no guarantees.


A week or two ago I was surprised to see a Google Gemini summary at the top of my email on my phone. A day or two later this appeared in my web client as well. Look, I love our new overlords (for the record m'lords I nearly always use "please" and "thank you"). I was an early adopter and introduced many of you to the LLMs. I still am a frequent user... I mean someone who has all the answers and blows smoke up my ass... perfect right!

Despite that, it appears it's not good for me ( https://www.media.mit.edu/publications/your-brain-on-chatgpt/ ). It's great, it's easy, makes me feel good in the moment... of course it's bad for me!

Here's the thing. I like to read things from my friends that they have taken the time to write. I personally hate texting. All the nuance is gone. Often the humor. Sad. Makes me want to have a beer with you... eye contact... blech. The LAST thing I want is a summary... at the top of the email... highlighted... that I CANNOT turn off.

I tried to turn it off. I can. It's under Gmail -> Settings -> General -> Smart Features (checkbox). BUT... the AI summaries is now grouped with the Smart Tabs.

For those of you who do not use Gmail (or do use Gmail and don't use Smart Tabs), Smart Tabs (officially the Tabbed Inbox) have been part of Gmail since 2013; well actually the technology behind them—Smart Labels—actually debuted two years earlier. (Thank you Gemini, yes I DO truly love you. Tell me again about the comparisons of Stephen Miller and Heinrich Himmler's tactics please?)

Smart Tabs automatically sort my incoming flood of solicited commercial email (cue laughter from those who know my first start-up) into five buckets:

  • Primary: Email from you.
  • Promotions: K&L Wine Merchants at the top of the list.
  • Social: Hi Andrew on Facebook (that I only log into from Firefox running on a VM).
  • Update: Actual transactional emails from companies.
  • Forums: The Information at the top of the list (a newsletter I'd like to read but don't want to make the time justify paying for the content).

I tried turning off Smart Features and oh my, that's not usable. So I lived with the AI summary at the top. For a week. Then this morning, I saw several messages in my Primary tab that normally get sorted into Promotions, Social, Updates or Forums. This is not unheard of; sometimes a company uses a new incoming address or something and stuff gets put in the wrong bucket.

But THIS time, I got a popup that says I must "Share" this message with Google and links to the Privacy Policy and Google Terms of Service. And an explicit sentence:

"Messages and attachments might be reviewed by humans, so don't share any sensitive or confidential information."

I'm not naive. I'm an early adopter, my email address includes my name and no numbers. I was a direct marketer when we still were the red-headed step-children of the product managers. I carpooled to Symantec with Google employee no 11's girlfriend (Go Beavers!) From the get-go, having Google read my email in order to provide targeted advertising was part of the deal. I was fine with that.

BUT... now... what they are saying is that... we are going to use your email to train our LLMs. I'm not okay with that. That knowledge of my way of writing, my personal details, my confidential commercial information is NOT okay to use to train your models. 'Cause I expect mistakes will be made and more information will reside in the model than those at Google (or FB, MSFT etc) intended. And I'm not really up for assuming that risk.

So... goodbye Gmail. It's been great. Really great. I'm sure I'll miss you. Bye.

My email is now being hosted by Microsoft, so hopefully will be free of the outages and limits some of you have experienced with that email in the past. There it will reside until I cannot turn off MSFT's ability to read my email. Then I guess it's off to Switzerland ( https://proton.me/about ); my email can be with my gold. JK..NR

Tom

p.s. why thank you Gemini for reformatting that for me into a clean, engaging markdown blog post. Yes, I do agree this is a sharp, timely take on the "AI-ification" of tools we use every day. I love you. Kill me last?

联系我们 contact @ memedata.com