Exclusive-Meta testing a 'human concierge' for its new personal AI agent Muse

FILE PHOTO: People are seen behind a logo of Meta Platforms, during a conference in Mumbai, India, September 20, 2023. REUTERS/Francis Mascarenhas/File Photo

NEW YORK, Sept 22 (Reuters) - Meta ⁠has been testing a “human concierge” for its new personal AI assistant, Muse, which entails having human contractors ⁠quietly handle some of the phone calls placed via the digital agent, according to internal company posts ‌seen by Reuters.

The Facebook and Instagram owner told employees about the test last week, shortly after the company publicly launched a phone calling feature for Muse.

Some employees have raised privacy concerns about human agents handling the calls, warning that it could result in sensitive information being shared unintentionally with ​contractors in call centers, the posts showed.

Asked about the feature, a Meta ⁠spokesperson said the response from employees has been "overwhelmingly ⁠positive" and that the purpose of the test was to "get feedback so we can implement safety and privacy protections ⁠and ‌improve features before we release them publicly."

"We're working with merchants to continue improving this potential calling feature, and will only roll it out when it's ready and with the proper disclosures," said the spokesperson, Daniel Roberts.

MUSE TOPS ⁠USDOWNLOAD CHARTS

Meta's Muse agent has topped US app download charts in the ​two weeks since its debut, with ‌more than 2.5 million downloads, according to data provided by market intelligence firm Sensor Tower.

The centerpiece of CEO ⁠Mark Zuckerberg's plan to ​supply "personal superintelligence" to the billions of people who use Meta's services daily, the AI bot can autonomously perform tasks like sending emails, shopping and booking travel on a person's behalf.

Meta heavily emphasized the AI assistant's safety protections during its launch, telling would-be users that ⁠each agent would have its own secure “virtual machine,” or a dedicated ​computer in the cloud designed to keep people’s sessions isolated, and would keep sensitive information like passwords in separate secure storage.

With the phone calling feature, a Muse user can tell the agent to dial phone numbers for USbusinesses and talk with people ⁠on the other end to handle errands like booking haircuts, checking whether a store has an item in stock and getting quotes from different contractors.

After the call, the agent provides a summary and transcript of how the exchange played out in the Muse app.

Meta began having employees test out, or “dogfood,” the phone calling capability in August and gradually started rolling it ​out to Muse users in the days after the app’s debut, the internal posts ⁠showed.

It enabled the human concierge feature, also referred to as “human agent calls,” for half of Meta’s employees last week, according ​to the post announcing the feature to employees.

“Muse is now able to hand ‌requests to a trained agent, who places the call and ​works it through,” the post said.

Employees who did not want to be included could join an opt-out group, according to the post.

(Reporting by Katie Paul in New York, Editing by Kenneth Li and Matthew Lewis)

Thank you for your report!

Sum Technology to acquire 51% stake in INI Technologies for RM2mil

Priority Technology to go private in $1.6 billion CEO-led deal

UN experts: AI safety measures failing to keep pace with technology

资料照片：2023年9月20日，在印度孟买举行的一场会议上，人们站在Meta Platforms的标志后面。路透社/Francis Mascarenhas/资料照片

纽约，9 月 22 日（路透社）——据路透社看到的 Meta 公司内部帖子显示，该公司一直在为其新的个人人工智能助手 Muse 测试“人工管家”服务，这意味着由人工承包商“悄悄地处理一些通过数字代理拨打的电话”。

Facebook 和 Instagram 的所有者上周向员工透露了这项测试，此前不久，该公司刚刚公开推出了 Muse 的通话功能。

帖子显示，一些员工对人工客服处理电话的隐私问题表示担忧，并警告说这可能会导致敏感信息在无意中与呼叫中心的承包商共享。

当被问及该功能时，Meta 的一位发言人表示，员工的反应“非常积极”，测试的目的是“获得反馈，以便我们能够在公开发布之前实施安全和隐私保护措施并改进功能”。

发言人丹尼尔·罗伯茨表示：“我们正在与商家合作，继续改进这项潜在的通话功能，只有在准备就绪并做出适当披露后才会推出。”

MUSE 登上美国下载排行榜榜首

根据市场情报公司 Sensor Tower 提供的数据，Meta 的 Muse 代理程序自推出以来两周内一直位居美国应用程序下载排行榜榜首，下载量超过 250 万次。

作为首席执行官马克·扎克伯格计划的核心，该人工智能机器人旨在为每天使用 Meta 服务的数十亿人提供“个人超级智能”，它可以自主执行诸如发送电子邮件、购物和代表用户预订旅行等任务。

Meta 在发布会上着重强调了 AI 助手的安全保护措施，告诉潜在用户，每个代理都将拥有自己的安全“虚拟机”，或者说是云端的一台专用计算机，旨在隔离用户的会话，并将密码等敏感信息保存在单独的安全存储中。

借助电话功能，Muse 用户可以告诉代理拨打美国企业的电话号码，并与另一端的人交谈，处理诸如预约理发、查询商店是否有库存商品以及从不同的承包商处获取报价等琐事。

通话结束后，经纪人会在 Muse 应用中提供通话过程的摘要和文字记录。

内部帖子显示，Meta 从 8 月份开始让员工测试（或称“内部测试”）电话功能，并在该应用程序推出后的几天内逐步将其推广给 Muse 用户。

据向员工宣布该功能的帖子称，上周 Meta 为一半的员工启用了人工客服功能，也称为“人工代理电话”。

帖子中写道：“Muse现在可以将请求转交给训练有素的代理人，由代理人拨打电话并跟进处理。”

根据帖子内容，不想被纳入其中的员工可以加入退出小组。

（纽约 Katie Paul 报道，Kenneth Li 和 Matthew Lewis 编辑）

感谢您的报告！

Sum Technology将以200万令吉收购INI Technologies 51%的股份。

Priority Technology将以16亿美元的价格私有化，该交易由首席执行官主导。

联合国专家：人工智能安全措施未能跟上技术发展的步伐