Do People Trust Conversational AI?

BY • POSTED June 17, 2019
Home / Uncategorized / Do People Trust Conversational AI?

New data suggests that people, including small business owners, don’t yet trust artificial intelligence (AI) to make phone calls or send emails on their behalf.

Nearly 73% of people are unlikely to trust conversational AI such as Google Duplex. Google Duplex is an AI-powered voice assistant that can call and book restaurant reservations for smartphone users, according to a new survey by Clutch.

Most people (70%) also say they are unlikely to trust AI to reply to simple emails for them.

Conversational AI enables people and computers to communicate in a way that imitates natural human conversation. These tools can benefit businesses, but only once people come to trust them. Experts say that consumers will embrace conversational AI when it becomes more familiar, as they have with emerging technologies of the past.

This article explores how people currently view conversational AI, how that may change in the future, and how this technology benefits businesses.

Duplex Requires Users to Hand Over Too Much Responsibility

Artificial intelligence adds convenience to communication but, as Clutch’s data indicates, many people don’t yet trust an AI tool to speak for them.

Experts in the survey’s report suggested that users like to monitor and control an AI-powered interaction. A phone call may offer too little capabilities for monitoring. For example, Google Duplex users can delegate the task of making restaurant reservations, but most people consider this to be too much responsibility to hand over to a computer.

People worry that AI-powered calling can be subject to errors, both in relaying information correctly and in ensuring that information is received accurately. These users don’t want to show up to a crowded restaurant with an inaccurate reservation.

Without an interface to catch mistakes, people must trust that the AI has successfully completed the task. New technologies are often perceived as unreliable, and, in turn, avoided until the perception is corrected. At present, the fear of failure outweighs the benefit of saving time.

Google itself may realize that the AI isn’t yet advanced enough to escape errors. In May 2019, it was revealed that up to 25% of Google Duplex’s calls are actually placed by humans. These human-powered calls will help train Duplex to advance in the future.

As voice-powered AI becomes more widespread, though, trust for these tools is expected to increase, especially if it proves it can complete tasks with few mistakes.

People Unnerved By Chance of Mistaking AI for a Human

People want to be able to differentiate between humans and machines. According to the report, 61% of people would feel uncomfortable if they believed they spoke to a human and later learned they had spoken to AI. After all, it’s scary to think that a computer could ever be indistinguishable from a human.

It also makes people uncomfortable to see technology overtake a function that was previously performed by humans, said Ivan Kotiuchyi, a research engineer at Ciklum, a software development company. “We’re all vulnerable to different forms of this anxiety when it comes to adjusting to new ideas or technological advances,” said Kotiuchyi.

For example, both ATMs and self-checkout lines received pushback from consumers until the tools proved themselves to be at once valuable and mundane. In a similar fashion, people’s apprehension toward AI will wane as the technology becomes more commonplace.

People Want AI to Announce Itself

People want AI to declare itself on the phone, as it can be used for malicious purposes. Robocallers already use bots to mimic humans in order to scam consumers. Conversational AI could be used to further manipulate people, allowing bots to perfectly match the vocal patterns of a victim’s boss, family member, or a government agent.

This fear of manipulation may be why 81% of people want an AI-powered voice assistant such as Google Duplex to declare itself as a robot before proceeding with a call.

There currently isn’t much regulation indicating if AI needs to declare itself. In July 2019, though, California will begin requiring AI-powered messages to announce they are from a bot. While such protections will help consumers to feel more at ease when interacting with AI, it does little to undo the threat of manipulation.

People Hesitate to Let AI Send Their Emails

Lastly, nearly three-quarters of people (70%) say they are unlikely to trust an AI-powered assistant to reply correctly to simple email messages. And yet, conversational AI is already used widely in written communications, most notably in the case of Gmail’s Smart Reply feature. Smart Reply is now responsible for 1 in 10 emails sent by the Gmail App, yet most people don’t recognize this as AI.

This is likely because the email interface presents users with a set of choices, then lets them click ‘Send’. Since people feel in control of this interaction, they’re less likely to realize or react negatively to the influence of AI.

Conversational AI Can Help Businesses

Despite consumers’ lack of trust, conversational AI has the ability to benefit businesses in the future. Conversational AI enables businesses and their call center companies interact with customers more efficiently. It can add personalization to automated processes, allowing businesses to maintain the customer experience while still saving time and money on resources.

People prefer to speak to a human when calling customer service, so human-sounding voice automation could be used to improve the call experience for customers. Conversational AI can be used to identify the caller, their needs, and their mood before transferring them to a human service representative.

This way, businesses can improve their efficiency and cut costs, but also introduce a human touch that improves the overall customer experience.

Familiarity with AI Leads to Trust

People don’t yet trust conversational AI to communicate on their behalf, but that could soon change. As this technology evolves, businesses can expect consumers to be more comfortable with AI-driven communication, whether on the phone or by email.

As people become more tolerant of bots, AI-supported call centers and chatbots can help your business to better engage, support, and retain customers.

Author Bio:

Riley Panko is Senior Content Developer for Clutch, the leading research and reviews platform for B2B marketing and technology services. She focuses on business and answering services research.

Schedule a tour by clicking on the location below:

Highland Heights
Lakewood
CoWork Oberlin: Powered by LaunchHouse

Sign up for our newsletter and get a free day pass to work out of our space! Plus you'll get exclusive membership deals and events updates.