Will AI Assistants Save Us From Administrative Tedium?

37 percent of survey respondents cited administrative work, such as scheduling meetings and data input, as tasks that would be outsourced to machines or automated by 2030. What do these AI assistants look like, and how do they work?

By David Ryan Polgar, Contributor

Workers spend a lot of time doing tedious administrative tasks, such as scheduling meetings, collecting and transferring data within a company, and creating reports around performance. A UK survey last year found that employees spend up to a third of their working day on administrative or repetitive tasks. Another survey of American and UK companies found that managers spend 40 percent of their time doing administrative tasks, which may be getting in the way of focusing on strategic work.

Yet, there may be an alternative digital solution on the horizon. In an interview with Nasdaq Disruptors, founder of x.ai, Dennis Mortensen predicted a future in which cloud-based voice services like Siri or Alexa could satisfy broad assistance while a crop of specific-task AI would take over administrative tasks.

Business leaders are starting to agree. In the Dell Technologies Realizing 2030 research results, 37 percent of respondents cited administrative work, such as scheduling meetings and data input, as tasks that would be outsourced to machines or automated by 2030.

In fact, that time may already be here.

37 percent of survey respondents cited administrative work – such as scheduling meetings and data input – as tasks that would be outsourced to machines or automated by 2030.
— Dell Technologies Realizing 2030 research

Meet Your New Digital Assistant

NYC-based x.ai, an AI technology company, developed Amy and Andrew Ingram—anonymous AI personal assistants—who do the tedious work of scheduling for you. After copying Amy or Andrew into an email, the intelligence assistant will take over scheduling. (Amy and Andrew can also be used within Slack.)

As it turns out, Andrew and Amy are part of a rising trend to use AI to help employees solve everyday problems.

According to Keith Johnson, CTO at Fuze, a cloud-based unified communications platform that focuses on productivity, the development of virtual assistants will mirror the personalization of our current productivity tools, whereby employees will pick and choose tools depending on preference and need.

“[In] 2018, we see the first widespread practical uses for AI-enabled bots to automate a number of business tasks, including scheduling,” Johnson stated.

Instead of relying on a single software suite deployed to the whole company, Johnson is banking on a world whereby the most successful organizations let employees leverage a host of more exact solutions. “[W]e’re going to see an explosion of smaller scale AI-powered productivity tools that cater to a specific function of your day – sorting emails, scheduling meetings, and so on.”

Another example is Leo, a robotic assistant that automates tasks such as data transfers, form completion, and financial auditing—in essence, time-consuming tasks that are monotonous in nature. A professional working in HR, for example, could use Leo to quickly pull the social media information of a stack of applicants and have it placed into an Excel spreadsheet for hiring managers to easily assess.

Developed by Kryon Systems, Leo can work 24 hours a day. But in order for companies to adopt AI-based virtual assistants, both employees and business leaders need to begin to trust in the accuracy and efficiency of these types of tools.

The key for business leaders in 2018, co-founder of Variable Labs Adam Zuckerman said, is to ensure AI bots add value in a way that doesn’t require an overhaul in the ways things currently operate. “AI/bots will make business more efficient, but it’ll only do so when it reaches the point of adding more value or can operate without significant change by groups using it,” he said.

Adding Value, Not Workload

The Nikabot was built by the London-based company Impossible Labs to simplify and automate administrative tasks. According to Victoria Ivanova, who is part of the team who built Nikabot, the bot was initially built for internal use to mimic (and take some of the workload off of) their studio manager, Nika.

At the time, the human Nika was gathering the employees’ time sheets and reporting their activity to the company’s CFO to bill clients and pay employees. For the bot version on Slack, the Nikabot asked one question: “What did you do today?” The slackbot then collated that information to speed the billing process.

According to Impossible Labs, Nikabot is now used by over a thousand businesses to automate the process of tracking what its workers are doing throughout the day. For businesses that have a significant remote workforce, this may be particularly helpful. Nikabot collects the data from workers and automatically generates reports that may provide useful insight around where employees are spending their time.

Every organization has a lot of low-level, shallow work—scheduling meetings, gathering data for reports, team-wide surveys, sales calls [the list goes on]” Ivanova said. “If there is a possibility for automation, then it should be done so that people are left to work on more complex, value-generating tasks.”

Whether that automation will be done entirely with bots she went on, is hard to say. But, what is clear for Ivanova? “Automation itself, paired with machine learning and AI, is practically inevitable.”

Moving Toward an ‘AI High-Trust’ Society

Berlin-based data science anthropologist Ali Rebaie divides the different work-culture environments into two categories: “AI high-trust” and “AI low-trust cultures.”

“As we move from a task-oriented society towards automation, an AI high-trust culture can organize its workplace [around] a more data-driven basis, with more responsibility from top management delegated to virtual assistants,” Rebaie stated.

In contrast, according to Rebaie, AI low-trust cultures will be slower to adopt virtual assistants because they are applying the traditional rules of the workforce regarding chain-of-command and information control. “Trust tremendously affects delegation,” Rebaie stated. “It will help facilitate the cooperation with virtual assistants and reduce the necessity of monitoring.”

Virtual assistants like x.ai, Leo, and Nikabot offer the potential to free up burdensome administrative tasks. An unanswered question, however, is whether businesses have enough trust in AI assistants in order to delegate the work.