Sources: Pizza Hut, Staples, Whole Foods
Do chatbots need to be more likable?
A new university study finds people will more readily use a chatbot if they perceive it to be friendly and competent and less so if it projects overconfidence and arrogance. The challenge, the authors say, is finding the right balance.
Across three studies with 300 participants in the U.S., researchers tested reactions to AI-bots with the same underlying functionality but different descriptions.
Among the findings:
- Low-competence descriptions (e.g., “this agent is like a toddler”) led to increases in perceived usability, intention to adopt and desire to cooperate relative to high-competence descriptions (e.g., “this agent is trained like a professional”).
- People are more likely to cooperate with and help an agent that projects higher warmth (e.g., “good-natured” or “sincere”).
- Descriptions “are powerful,” helping drive user adoption and engagement with chatbots.
The authors suggested chatbots need to instill confidence that they are worthwhile to engage with. At the same time, acknowledging some errors may occur early on as the chatbot learns what users want will likely help people become more accepting of a chatbot’s mistakes. Pranav Khadpe, a co-author, told The Wall Street Journal, “You really want to manage the expectations you set before the first interaction.”
Amazon.com’s Alexa and Google Assistant have significantly expanded their features to make conversational flow more natural in order to encourage engagement. Adding jokes and clever responses to certain questions are some ways voice assistants are working to elevate the experience with more personality.
According to an Alexa developers page, the five personality characteristics that “make Alexa, Alexa” are: approachable, efficient, trustworthy, natural, and responsive.
An article last year from CNET explored how improving two-way interactions was seen as key to helping voice assistant users graduate to more advanced activities like shopping or controlling other smart devices in the home.
“A voice-based assistant must be more intelligent, proactive and understanding than a search results page,” Kristen Hanich, a research analyst at Parks Associates, a smart home market research firm, told CNET. “It is not enough to present possible answers to a voice query; rather, assistants will need to present the right answer in a personal and meaningful way.”
- How to Build a Likable Chatbot – Stanford University
- Conceptual Metaphors Impact Perceptions of Human-AI Collaboration – ACM (Association For Computing Machinery)
- Consumers Like Chatbots to Be Smart—but Not Too Smart – The Wall Street Journal
- The Alexa Personality – Amazon Alexa Developer
- Alexa and Google Assistant are developing personalities – CNET
- Prepare for the voice revolution – PWC
- Voice Trends & Statistics: What Designers Need to Know About the New Tech Boom – Adobe
- Does humanizing virtual assistants undermine consumer privacy? – RetailWire
- COVID-19 may push retailers to use voice assistants instead of touch screens – RetailWire
Discussion Questions
DISCUSSION QUESTIONS: What will make households more comfortable using voice assistants for shopping, controlling smart devices and other advanced features? What do you find most annoying and most useful about chatbots?
Poll
BrainTrust
Cathy Hotka
Principal, Cathy Hotka & Associates
Liz Crawford
VP Planning, TPN Retail
In general, people want a friendly and easy experience. It doesn’t matter if it’s human-to-human or human-to-AI/chatbot. It must be friendly and easy. So, let’s start there. It’s easy to program the chatbot to use the customer’s name and say, “Good morning,” or “Thank you.” That’s where it starts.
Most annoying is when the chatbot doesn’t understand the conversation and the system doesn’t have an easy way to get the customer to a live agent.
Usability is the key here. If a chatbot does the job simply and effectively then it is useful and provides benefit to the customer. If it doesn’t do that then it is frustrating. I also think there should be an option for the customer to speak to a human. At the end of the day, as much as technology has advanced, not everyone wants to speak with a chatbot in all circumstances. Giving users options is vital.
I have not had good experiences with chatbots. I never have gotten to where I need to go using a chatbot, so I use email which will hopefully get to a human that will answer my question.
Chatbots are still a work in progress. They’ll come along, but the most important success factor still is to transfer customers to a human agent when needed.
I like humans. And humans also need jobs. I find talking to a chatbot annoying, period.
I love technology but not this! I HATE it when you call customer service, knowing you need a person and you get six levels of press 1 for this and 2 for that. Chatbots are just another manifestation of the marketer saying “I care more about efficiency of my staff than I care about your time and frustration, my customer.”
While it is important for chatbots to have a likable personality, it will be great if they can mimic a real person on the shop floor. For example, through sentiment analysis, they can get a pulse on the customer’s mood to adapt their approach accordingly; or make small talk based on customer profile, location, etc.
While most of the chatbots out there are not known for distinct personalities like Alexa or Siri, people do like to interact with the kind of people they relate to. Subconsciously, we see a chatbot as another person. But brands would do well to remember that a chatbot on the website is a representation of their brand voice.
So if the brand is positioned as an expert on a serious subject and the typical audience is informed – Think a B2B professional services company – a chatbot can’t look like a toddler or give irrelevant answers. On the other hand, we do expect a chatbot on a Disney website to be different and relatable.
I find chatbots to be helpful for most questions and easy transactions. While I get frustrated sometimes when Alexa can’t seem to find the right answer despite repeated attempts, by and large I do like the technology for answering quick questions or putting things on a to-do list right when I think of them. I also find them very useful for reminders during my day, notifications of shipments and task oriented events. For shopping I’ve found them to be far less useful. Unless I know exactly the item I want, these tools are really not useful at all. Browsing items verbally is really tedious.
Chatbots could be useful if the experience is intuitive and ultimately leads to the customer’s request or problem being solved. Removing the frustration and friction out of the experience and being solution-driven makes the chatbot model effective. Context is absolutely critical, and if the chatbot leads to an endless loop of questions without solving the problem, then where is the value add of that experience?
If the experience is not what the customer expects, they may default to speaking to a live customer service representative to solve the problem. While efficiency matters in filtering out customer requests, the experience should be personalized to the extent that the chatbot “knows” who you are and has the context behind your request to solve the need. Otherwise, it’s going to lead to a friction-filled experience and plenty of customer frustration.
I’m with the majority here, these things are just an annoying ploy to avoid putting a real caring human on the phone to help you deal with a friction point. They’re right on par with the offshore support person that can’t speak English and says your name over and over again as a scripted ploy to make you think they are competent and supportive. Customers are so over this stuff and just want brands to show that they care as much as they say that they do — not throw up a million barriers to get the help that you need in an effort to get you to hang up and go away. I don’t mean to be negative here, it’s just not that hard to give people the help that they need if you really care.
Chatbots need to be more useful, not more friendly, and the biggest challenge is that shoppers don’t have an easy command taxonomy to interact in helpful ways. One of the key pieces of chatbot design is to not let the consumer think it is a human, so expectations are different. The key hurdle is usefulness over frustration.
Chatbots are just fine until they are not. It is as if their #1 objective is to make sure you never talk to a human rather than to be a helper in any way.
Chatting with an inane chatbot feels like an insult. Obviously the company doesn’t want to pay customer service people. All of the happy talk in the world doesn’t make it less demeaning.
Let me know if you’ve ever had a phone call that sounded something like this. “A human. I want to speak with a human.” Luckily AI doesn’t take irritation and even namecalling personally. If it did, neither Alexa nor Siri would be taking queries from many people I know.
Having authored a whitepaper on chatbot persona and worked deeply with bots and artificial assistants using both textual and voice interfaces, I can guarantee that it’s easy to say what the bot/AA should do and quite hard to make it happen.
For example, in order to create a better user experience, a great human agent can perceive a customer’s tone/persona and adapt to it in the moment. Many, human reps however, are poor judges or act the same no matter what, trudging through their work robotically. And as happened to me just a week ago with a Home Depot rep that was both rude and clueless, can cost their employer sales.
AI technology is making inroads to being a “properly” reactive respondent but what proper is very much depends on the perception of the user it’s interacting with, just as with human reps. But unlike human reps, there is only one AI system with one intelligence. So given the need to represent a brand identity without offending users and without damaging the brand and the state of technology presently, most bot deployments will err on the cautious side and not lean towards a persona that can be offensive.
As far as understanding user wants and utterances, NLU/AI training improves constantly just as e-commerce did to get us to where we are today. It’s a separate issue from persona that some may not understand well.
While I understand the need for them, I am not a fan of chatbots. If an issues is at all complex my experiences is they are useless at best and extremely aggravating at worst. I get tired of conversations where in spite of my continuing to say agent they keep going around in a circle. I always prefer to speak with a person.
We have to look at the chatbots and their underlying AI, particularly NLP as “work in progress.” If we move forward 10 even maybe 5 years, we will be surprised where we started from. These things are getting better, understand the content and tonality, and connect seamlessly to the databases already. They aren’t exactly replacing all humans but they will replace your “let me call my superior” workforce pretty rapidly. Also, they are inexpensive which means they are going to be deployed everywhere and we will have to live in a society where we talk to them and get used to them.
Consumers are already using voice assistants for simple tasks, they will simply evolve into more complex interactions. I expect chatbots to be just another way to interact with a database, but I believe they can be so much more if personalized with loyalty/club data. It stops being a chatbot and becomes MY chatbot — a more congenial proposition.