Photo: @Asheesh via Twenty20
Are AI-powered customer service agents already human enough?
A new university study finds that what matters in the customer service experience is the perceived humanness of the interaction and that often an agent powered by artificial intelligence (AI) is as good or better than a real person.
The study from University of Florida’s College of Journalism and Communications asked participants to chat with bots or human agents from companies like Express, Amazon.com and Best Buy and rate them on “humanness.”
Of the 172 participants, 63 (or 37 percent), were unable to identify whether they were interacting with a human or a machine.
Regardless, whether AI-driven or not, higher scores of perceived humanness led to greater consumer trust in the companies.
“If people felt like, if it was human — either with really good AI or with a real person — then they felt like the organization was investing in the relationship,” said Tom Kelleher, a University of Florida advertising professor, in a statement.
The bigger issue is the language used. The authors noted that past research has found that consumer trust and satisfaction grew as human customer service agents have increasingly used less “stuffy, stilted language,” and the same holds true for chatbots.
Indeed, AI-powered agents aren’t as prone to become tired or jaded by routine interactions as human ones. Prof. Kelleher explains, “An agent can be so scripted that people feel like they’re talking to a machine.”
Past surveys have found people overwhelmingly prefer human assistance but value chatbots for their 24/7 availability and quick response. A Usabilla survey from 2018 found over half (54 percent) of respondents would always choose a chatbot over a human customer service rep if it saved them 10 minutes. Chatbots are also preferred for simpler tasks, such as changing or canceling orders.
AI use for customer service is expected to expand as people get accustomed to chatbots and AI functionality improves.
Ross Daniels, chief marketing officer at Calabrio, a workforce management software firm, told CMS Wire last year, “From providing near real-time feedback on customer and employee insights and stress predictors to voice-of-the-customer innovations, AI-driven analytics can quickly analyze the sentiment of customer interactions equipping contact centers with the visibility they need to help optimize the customer experience.”
- Chatbots or real people? Study finds customers only care about ‘perceived humanness’ – University of Florida
- STUDY: Over Half of Customers Prefer Speaking with Human Customer Service Agents via Phone – Usabilla/PRWeb
- What do your customers actually think about chatbots? – Userlike
- Is 2021 the Year AI Dominates the Call Center? – CSM Wire
BrainTrust
Neil Saunders
Managing Director, GlobalData
Cathy Hotka
Principal, Cathy Hotka & Associates
Gene Detroyer
Professor, International Business, Guizhou University of Finance & Economics and University of Sanya, China.
Discussion Questions
DISCUSSION QUESTIONS: Do you see a point in the near future when AI-powered customer service agents will consistently bring more value to the customer resolution experience than humans? Will they be preferred for the majority of tasks in the near future?
No.
Well said!
What matters most is solving a problem or answering a query in a satisfactory way. To most people, it matters little how this is done so long as it gets done. However the challenge for AI is two-fold. Firstly, AI is not so good a resolving complex or unusual queries. Second, it doesn’t have much of a personality which is not so great for brands. A good, personal interaction with a service agent strengthens the image of a brand.
Absolutely YES. The simple fact is, the sophistication in some AI-powered customer service agents enables them to actually do a better, more consistent job than human operators can. Ultimately, people that call customer service need answers. If chatbots can get answers faster, then customers will have no issues using them. But that won’t matter much, because as we go forward people will hardly be able to tell the difference between humans and chatbots.
Seeking “human-ness” often interferes with solving problems. Fake human feeling is not going to be a good long term strategy. Customers will be most satisfied when their problems are solved.
There’s a persistent but mistaken impression that AI will eliminate the need for humans. It’s true that chatbots can present as infinitely patient (I know I cannot), and can provide a friendly interface to customer interactions for many CS issues. And some, but not all, people actually prefer to interact with an anonymous machine rather than an actual person. However AI still cannot replace humans in many customer interactions – from complex information requests to issues of judgement. The best way to think of AI, or technology in general, is as a human augmenter rather than as a human replacement.
I agree with Neil Saunders – it’s whether the problem gets solved quickly and easily. Do that and nobody cares whether it’s live or AI.
Chatbots can sound human, but the trust will improve if it is disclosed up front that customers are dealing with bots and help is only a tap away.
Effective utilization of chatbots in improving customer service outcomes depends on three aspects.
I just got off the phone with an AI agent. The answer is definitely NO.
Humans are multi-dimensional, and the same applies to customer service needs. An AI-powered customer service agent will get you only so far with a typical customer service request. If customers follow a linear “happy path” for their customer service requests, an AI-powered customer service agent may be able to address their needs.
However the typical customer service call is ever on the “happy path,” as every need and ask has a unique journey, and this is where the AI-powered agents fail in most cases. There is still significant work to be done before we could even consider AI-powered customer service agents a viable option.
AI will replace humans for rotes task and if customer service tasks are rote then yes, chatbots will be preferred, and humans can be upskilled for such things as providing insights from the information AI gathers. This was nicely articulated by Ross Daniels: real-time feedback on customer and employee insights, stress predictors, and voice of the customer.
Interacting with a human agent who is friendly and solves your problem quickly is a very satisfying experience. A human agent who is incompetent generates a number of four-letter words during the dialogue.
Net/net — solve my problem, retailers! I really don’t care if it is an AI or human who solves it.
I always prefer a live human being for customer service, however in a 24/7 world, the area where chatbots can be highly effective is off-hours where a consumer can give/get information about an order or a situation such that it moves the needle. The key to this is PROGRESS. If a consumer doesn’t see progress in moving the issue towards resolution, then who cares about the improvement of “stuffy, stilted language?” AI chatbots won’t replace highly important customer service professionals.
Anyone who has been trapped in “voicemail jail” will say that we’re not there yet. Things are getting better, but it’s not the time to declare victory.
I’m surprised we’re not already there. It feels like every time open a chat on an e-commerce site I find myself chatting with a bot. For simple routine stuff, makes sense and they work. When there are issues or I need something that’s not 100 percent clear, then it’s a not such a great experience. Once you get trapped in a loop with one of these agents, it gets frustrating. I do believe that they’ll get better, but right now — we’re not there yet.
AI brings value for the lower level, easier-to-answer, common questions and for managing repetitive tasks. The key is to simulate the human-to-human experience as much as possible. The best systems out there still can’t answer every question. Higher level questions and problems are still solved by human agents. The agents like that AI is supporting them in this way. It will be many years from now, if ever, before we might see AI managing all customer support issues.
There are multiple variables at play here. The first, obviously, is what problem is being solved. If it something simple and routine, who needs people and phone queues? Also there is the language issue. With so many call centers offshore the odds of finding a customer service agent sans accent decrease and many people act like American tourists on the phone screaming at the agent because they (the callers) aren’t paying enough attention to “hear” through the accent or because they resent jobs being transferred to another part of the world, or whatever. But there are some instances when irrational customers need their hands held, their egos affirmed, or just reassurance that their problem will be resolved.
This past Saturday I spent an hour trying to convince Amazon that I really live at East 11 Mile Road, and not West 11 Mile which is where they seem to send every other package. The AI system didn’t seem to understand. Of course, to be fair, neither did the human service agent. And I guess that’s it. For today at least, AI is fine for transactional matters, but you still need humans when you are dealing with relationship issues. As to the future the difference between AI and humans is that AI evolves, improves, and gets smarter over time.
Upon reading the opening sentence “what matters in the customer service experience is the PERCEIVED HUMANNESS of the interaction and that often an agent powered by artificial intelligence (AI) is as good or better than a real person,” the term “perceived humanness” really stands out, and I’m not sure it’s the right term. In this context I think it is probably defined as “a relatable, empathetic, and effective problem solver.” Not all humans have those qualities, but AI can be designed to do just that (whether real or a mirage), so in that sense – YES, AI will be able to bring more (and more consistent) value to some customer resolution experiences than humans.
AI is great for uninvolved issues, but when I say “agent” I want to speak to a person.
If the chatbot addresses the issue in a straightforward, simple and understandable way, it really doesn’t matter. However, there needs to be a flexibility built in that when AI “gets stuck,” a human needs to intervene.
Organizations are becoming more customer-centric. So, if AI-powered customer service agents add more value to the customer experience, that’s what businesses will want. I believe it is critical to focus on how new technological capabilities solve customer problems and improve their experience. This can help increase customer loyalty and, as a result, stakeholder confidence in the company. There is no doubt that AI-powered customer service agents will be preferred in the near future as long as they serve the aforementioned purpose.