Artificial Intelligence

Top 4 challenges to adopting conversation bots in the enterprise

While conversational agents are helpful in assisting teams and improving efficiency, they also come with obstacles.

One of the big ways artificial intelligence (AI) has gained ground in the home and the enterprise is through conversational agents, according to a Thursday report from KPMG. Using these agents—which the report defines as "chatbots and intelligent assistants that interact with people via voice or text channels, on devices such as smartphones, automotive infotainment consoles, and smart speakers"—has become second nature to consumers, who speak to them daily in their homes, cars, online shopping sessions, and more.

A further sign of the spread of conversational agents is smart speaker sales, which IDC predicts will hit nearly $12 billion by the end of 2018, and reach $28 billion in 2022, the report found. Additionally, AI applications are expected to increase from $644 million in 2016 to $37 billion by 2025, which is a factor of 56, the report found.

SEE: Electronic communication policy (Tech Pro Research)

Conversational agents aren't limited to consumers—these tools have a plethora of use cases in the enterprise, according to the report. Conversational agents could act as an office assistant, completing tasks, scheduling meetings, translating documents, ordering lunches, and more. They can also be used to streamline office operations, like scheduling conference rooms, or analyze data and draw conclusions, the report found.

However, organizations face several challenges when integrating conversational agents with their teams. The report identified the following four major obstacles conversational agents bring to the workplace:

1. Scarce training data. Even in very large organizations, an internal workforce will have a smaller user base, providing fewer interactions, than a bot deployed to the general public. This is particularly the case for bots built to automate niche functions.

2. Risk concerns. Even conversational agents deployed for internal consumption can introduce risk. In domains where the agent provides guidance, such as HR or Compliance, the organization may wish to review and vet all responses before bot deployment.

3. Accessibility. From a UX standpoint, the organization should also utilize inclusive design practices to ensure the bot interface is accessible to all potential users, regardless of working situation, capability, or language proficiency.

4. Privacy concerns. Just like customer-oriented virtual assistants, conversational agents in the workplace learn by collecting feedback and behavioral data. To manage privacy concerns, organizations should be transparent about how data may be used, and give employees an opportunity to choose to participate. After opting-in, employees should be able access their conversation history and have the ability to temporarily turn off the bot's listening capabilities if needed.

In order for conversational bots to successfully integrate into an organization, they must adopt these four foundational capabilities: They must learn, fail usefully, be personalized, and prioritize user experience, according to the report.

Businesses that want to integrate AI chatbots into their workflow must first see if their business goals align with the strengths of the technology. To learn more about how your company can benefit from chatbots, click here.

The big takeaways for tech leaders:

  • Conversational bots are intelligent assistants and chatbots that help people quickly accomplish tasks and are controlled by voice. — KPMG, 2018
  • These bots can be beneficial, but do bring about privacy concerns, risk threats, scarce training data, and limited accessibility. — KPMG, 2018

Also see

voiceassist.jpg
Image: iStockphoto/AndreyPopov

About Macy Bayern

Macy Bayern is an Associate Staff Writer for TechRepublic. A recent graduate from the University of Texas at Austin's Liberal Arts Honors Program, Macy covers tech news and trends.

Editor's Picks

Free Newsletters, In your Inbox