Ready to swap your old cube-mate for a disembodied AI? IPsoft CEO Chetan Dube, creator of AI co-worker AMELIA, gives his take on the post-COVID office landscape.

准备将您的旧立方体伙伴换成无形的AI？ AIsoft同事AMELIA的创始人IPsoft首席执行官Chetan Dube阐述了后COVID办公环境。

By S.C. Stuart

由 SC Stuart

The COVID-19 pandemic has emptied the world’s offices, sending most of us to work from home for the foreseeable future and facilitating a mass migration to the digital workplace. But will it also jumpstart the arrival of our non-human colleagues?

COVID-19大流行已使世界各地的办公室空荡荡，使我们大多数人可以在可预见的将来 在家工作，并促进了向数字工作场所的大规模迁移。 但这是否还会促使我们非人类同事的到来？

AMELIA (“AI + ME = AMELIA”) from IPsoft can read 300 pages in 30 seconds, comprehend multiple languages (including logic and context), and is installed in more than 500 banks, insurance companies, and retail giants.

IPsoft的AMELIA (“ AI + ME = AMELIA”)可以在30秒内阅读300页，理解多种语言(包括逻辑和上下文)，并安装在500多家银行，保险公司和零售巨头中。

PCMag Editor-in-Chief Dan Costa interviewed IPsoft CEO Chetan Dube in 2018, so we reached out to see what AMELIA is capable of in 2020, how it’s helping with COVID-19 recovery plans, and whether we’ll be welcomed back to our workspaces by AMELIA-like AI coworkers.

PCMag主编Dan Costa在2018年采访了 IPsoft首席执行官Chetan Dube，因此我们伸出援手了解了AMELIA在2020年的能力，它如何帮助COVID-19恢复计划以及是否会受到欢迎类似AMELIA的AI同事的办公空间。

PCMag: According to the World Economic Forum’s The Future of Jobs Report, half of the world’s employees will be digital by 2025. Chetan Dube: Accenture also published a report in November stating that up to 75 percent of organizations could go out of business if they fail to scale AI. When you think about the amount of jobs that would be lost in that instance, it is imperative that businesses embrace the technology. AI will help preserve organizations; it will allow them to persevere while freeing their human employees to focus on higher-value tasks. AMELIA is a real-world digital employee for today’s real-world challenges [and] on the forefront of this expansion, working alongside human employees, assisting them in a variety of ways to form the perfect hybrid workforce.

PCMag：根据世界经济论坛的 《就业的未来报告》 ，到2025年，全球一半的员工将是数字化的。Chetan Dube：埃森哲在11月还发布了一份报告 ，指出多达75％的组织可能会倒闭。无法扩展AI。 当您考虑在那种情况下将失去的工作量时，企业必须接受该技术。 人工智能将帮助保护组织； 这将使他们在坚持不懈的同时解放员工，使其专注于更高价值的任务。 AMELIA是一名现实世界的数字员工，在当今的现实世界挑战中[和]在此扩展的最前沿，与人类员工一起工作，以各种方式帮助他们形成完美的混合型劳动力。

At PCMag, we’ve already seen plenty of futuristic hybrid workforce experiments including: embodied cyborg-enabler exoskeletons which will give us enhanced strength to expert system AIs managing oil rigs and power plants, based on NASA technology and — further out — highly lifelike “synths” which mirror our movements, trained through emulation. AMELIA seems more pragmatic and already in place in many of today’s corporations. Give us some use case scenarios. Contact centers, for example, are overrun with inquiries that cannot be processed in a timely manner. This creates a lengthy queue of frustrated individuals who are stuck waiting for answers. Service desks are equally overwhelmed. Imagine working for a company with so many employees and so many issues that they’ve actually deployed multiple service desks. Who will you call when you have an issue? Do you know? Does anyone know the right number? And how many minutes or hours will you waste searching for answers?

在PCMag，我们已经看到了许多未来派的混合劳动力实验，其中包括：嵌入式 机器人启用外骨骼 ，这将使我们增强 基于NASA技术管理石油钻井平台和发电厂的 专家系统AI的 实力，并且-更进一步- 高度栩栩如生的“合成器” 反映了我们的动作，并通过仿真进行了训练。 AMELIA似乎更加务实，并且在当今的许多公司中已经存在。 给我们一些用例场景。 例如，联络中心因无法及时处理的查询而超支。 这造成了沮丧的人们排长队，他们被困在等待答案。 服务台也同样不知所措。 想象一下，在一家有这么多员工和很多问题的公司工作，以至于他们实际上已经部署了多个服务台。 遇到问题时，您会打电话给谁？ 你知道吗？ 有人知道正确的电话号码吗？ 您会浪费多少分钟或几个小时来寻找答案？

Is AMELIA also being used in ways we might not expect?Yes, for example, in serving as a front-end chat agent, she can also increase security by preventing malicious threat actors from socially engineering their way into customer accounts. She has done this for clients to great effect. She ignores the excuses and sob stories that hackers might share, which could persuade a human customer agent to divulge customer account details. AMELIA cannot be deceived like that. She will insist that the customer’s identity is verified before proceeding or sharing any personal information.

AMELIA是否也以我们意想不到的方式使用？ 是的，例如，在充当前端聊天代理时，她还可以通过防止恶意威胁行为者对进入客户帐户的方式进行社交设计来提高安全性。 她为客户做到了这一点，取得了巨大的成就。 她无视黑客可能会分享的借口和悲惨故事，这可能会说服人类客户代理泄露客户帐户详细信息 。 不能这样欺骗AMELIA。 她将坚持在继续或共享任何个人信息之前，必须先验证客户的身份。

Is AMELIA helping out on the frontline against today? Yes, our COVID-19 digital employee was built from the ground up to screen for potential indicators of the novel virus. Any member of the general public can talk to AMELIA, via voice call or browser, and receive an initial risk assessment based on their current symptoms and potential risk factors, drawing on recommended information from the US Centers for Disease Control and Prevention (CDC), as well as The World Health Organization (WHO), to ensure the information is always up to date. She also highlights where consumers can find more information about the pandemic.

AMELIA在今天的前线有帮助吗？ 是的，我们的COVID-19数字员工是从头开始构建的，目的是筛选新型病毒的潜在指标。 任何公众都可以通过语音呼叫或浏览器与AMELIA进行对话，并根据美国疾病控制与预防中心(CDC)的推荐信息，根据其当前症状和潜在危险因素进行初始风险评估，以及世界卫生组织(WHO)，以确保信息始终是最新的。 她还重点介绍了消费者在哪里可以找到有关大流行的更多信息。

Process memory allows her to dynamically navigate business process flows without the need to follow a step-by-step process to achieve a desired outcome. This is part of the reason why she can communicate so realistically. Much like a human employee, AMELIA also learns on the job by observing interactions between her human coworkers and customers. This allows her to independently build her own process map of what is happening. Last but not least, AMELIA’s affective memory uses state-of-the-art Affective Computing and Sentiment Analysis techniques to computationally model a user’s emotion, mood, and personality.

流程记忆使她能够动态导航业务流程，而无需遵循逐步的流程来达到预期的结果。 这是她能够如此现实地交流的部分原因。 就像人类员工一样，AMELIA还通过观察人类同事与客户之间的互动来学习工作。 这使她能够独立地建立自己的正在发生的过程的流程图。 最后但并非最不重要的一点是，AMELIA的情感记忆使用最先进的情感计算和情感分析技术对用户的情感，情绪和性格进行建模。

Does AMELIA use ELMo, developed by Allen NLP? AMELIA’s NLU (Natural Language Understanding) is based on ELMo (embedding for language models) and numerous other models, such as BERT and Electra, because of their functionality and flexibility. These models have allowed us to develop AMELIA as a digital employee that not only uses and recognizes individual words in a conversation but also understands the context through the surrounding words being spoken. As a result, less data is required to train the cognitive model. This makes AMELIA much more effective and easier to deploy for customers.

AMELIA是否使用 Allen NLP开发的ELMo ？ AMELIA的NLU(自然语言理解)基于ELMo(语言模型的嵌入)和许多其他模型(例如BERT和Electra)，因为它们具有功能性和灵活性。 这些模型使我们能够将AMELIA发展成为一名数字员工，不仅可以使用和识别对话中的单个单词，还可以通过周围的周围单词来理解上下文。 结果，需要较少的数据来训练认知模型。 这使AMELIA更加有效，更易于为客户部署。

When AI can create people that don’t exist, why was it important to you to model your AI on Lauren Hayes, a professional model? We believe that machines are not to be feared, but combined with humans, the hybrid workforce will make for a more efficient world. An effective way to foster this liaison is to humanize AI. We wanted consumers, who would often be the ones speaking to AMELIA, feel that she was as real as possible. When humans connect with AMELIA, they are more likely to speak to her and share more information. Thus, it made sense to base her initial design on a real person.

当AI可以创造出 不存在的人时 ，为什么对您 使用专业模型 Lauren Hayes进行 AI建模很重要 ？ 我们相信，不必担心机器，但与人类结合在一起，混合型劳动力将创造一个更加高效的世界。 促进这种联系的有效方法是使AI人性化。 我们希望经常与AMELIA交流的消费者感到她尽可能真实。 当人类与AMELIA建立联系时，他们更有可能与她说话并分享更多信息。 因此，将她的最初设计基于真实的人是有意义的。

In the spirit of background research, I had a chat with AMELIA on your site. The demo gave me a moniker (“John Smith”) and, I’m assuming, a “scenario” for AMELIA. This is how it went when I asked the AI to introduce me to its publicist so I could locate you:Do you have a publicist? [A] I do not, but I do have a team of writers, editors, and designers.Are you an AI? [A] No, I am not an AI. I am a real person. I work in publishing.Really? [AI] Yes, I work as a copywriter for a newspaper in New York City.How old are you? [A] I was created on December 25th, 1986. I was born in the United States.

本着研究背景的精神，我 在您的网站上 与AMELIA聊天 。 该演示给了我一个绰号(“约翰·史密斯”)，我假设这是AMELIA的“场景”。 当我要求AI将我介绍给其公关人员以便可以找到您时，情况就是这样：您有公关人员吗？ [A]我没有，但是我有一个作家，编辑和设计师团队。 您是AI吗？ [A]不，我不是AI。 我是一个真实的人。 我从事出版工作。 真？ [AI]是的，我是纽约市一家报纸的撰稿人。 你几岁？ [A]我是1986年12月25日创建的。我出生于美国。

During our conversation AMELIA clearly demonstrated sentiment analysis and affective computing skills, but I could see it needed a backstory. Clearly it’s not a bot designed to have a circular loop conversation. Right. She is pre-trained for roles and gets trained for others like her human colleagues. The difference is that-of course-AMELIA scales and provides results, which frees up her human colleagues to focus on high-value tasks.

在我们的交谈中，AMELIA清楚地展示了情感分析和情感计算技能，但我可以看到它需要一个背景故事。 显然，它不是旨在进行循环对话的机器人。 对。 她接受了角色的预培训，并接受了其他类似她的人类同事的培训。 不同之处在于，当然，AMELIA可以扩展并提供结果，从而使她的同事可以腾出时间专注于高价值的任务。

Finally, it’s clear that disembodied AI conversational systems will make up the bulk of the digital/human hybrid workforce, but is there a physical manifestation of AMELIA anywhere yet?Not yet.

最后，很明显，分散的AI对话系统将构成数字/人工混合劳动力的绝大部分，但是AMELIA的物理表现还没有出现吗？ 还没。

Originally published at https://www.pcmag.com.

最初发布在 https://www.pcmag.com 。