January 11, 2018 | Written by: David Ryan Polgar
Categorized: New Thinking
Share this post:
Question: Why do scuba divers always fall backwards out of the boat?
Answer: If they fell forward they would still still be in the boat.
Ba dum tsh. This dad joke comes courtesy of Microsoft’s Cortana. Chatbots, just like dads, want to be funny. As chatbots and other conversational interfaces become a part of our regular method of communication, it makes sense that we inject our digital helpers with a sense of humor to help them appear more human. According to a recent report by Twilio, 9 out of 10 consumers want to use messaging to communicate with brands. But how funny should these company or brand chatbots be?
Humor is subjective and ripe for missteps. One person’s laugh is another person’s cringe, as WIRED’s Brent Rose found out when he tried using the jokes from Cortana, Siri, and Alexa for a stand-up routine. More worrisome than falling flat and garnering groans, however, is offending your audience—which is always a risk with humor. What we find appropriate for laughter often depends on a shifting array of factors like tone, timing, and context. The complexity of what’s considered funny is enough to prompt many companies into developing bland personality-free chatbots. But creating a chatbot without humor would defeat the very purpose of most chatbots: creating a human-seeming conversation that people want to have.The trick, then, is finding a balance in creating a conversation that can build bonds and entertain users, but not offend or alienate.
“There has to be room for humor in chatbots,” states Anthony Kane, Sr., digital marketer and chatbot creator at 1SEO Digital Agency. The challenge, according to Kane, is creating an effective chatbot that has a soul. “The more we can make chatbots feel human the better,” he continues, while also providing a warning. “Until we see some major advancements in emotional intelligence the key will be to incorporate universal humor. When taking risks with pushing the limits of humor you open the doors to coming off as funny to one audience, but offensive to another.” “Universal Humor” is generally viewed as a baseline humor that doesn’t push the boundaries on sexual, political, or religiously-oriented jokes that could offend acceptable standards of certain communities or cultures.
In “The Role of Emotional Intelligence in AI”, founder of design firm MSTQ Yazin Akkawi emphasizes the importance of developing emotional intelligence in chatbots. He points to the common interactions we have with customer service agents, and the adjustment that is needed whether the person is angry, frustrated or impatient. In order to have a satisfying experience, the customer service agent needs to create a sense of empathy. That agent also needs to be able to toggle between different emotional states to mirror that of the other person. “For now, chatbots are becoming scary good at mimicking our language,” Akkawi wrote. “But until they can detect our emotional state and respond accordingly, they might never reach their full potential.”
As the editor of the Quartz Bot Studio, Emily Withrow builds conversational news experiences. She has spent a considerable amount of time thinking about what works and what doesn’t with chatbot experiences, and in a recent article confessed to spending more time with bots than people in a given week. I spoke with Withrow to get a better grasp of building a chatbot with personality, and inserting the correct level of humor in the conversation. Withrow served a previous stint as associate editor at the A.V. Club (The Onion), and in keeping with media stories of poets and artists creating chatbots, Withrow has a Masters in French Literature from Bryn Mawr College.
“Authors and performers have been conjuring up convincing artificial personalities for millennia, so it makes sense that they’d be on the vanguard of designing this new kind of software interaction,” wrote John Pavlus in a Fast Company article last year.
“We try to model chatbots on a good friendship,” says Withrow, speaking about the sometimes experimental chatbots created by the Quartz Bot Studio. “They are there when you want them to be, and they are not desperate.” A desperate, look-at-me-I’m-funny tone may be a turnoff for users. When developing conversational experiences, Withrow emphasizes that the humor is something that should organically unfold as opposed to seeming overly forced. “The humor is more subtle, more wry, and more built into the context,” she says.
This is a major point of distinction that appears to be missing in many of the chatbots I played around with. It is relatively easy to write a chatbot as a jokester, but that is likely a grating experience for the user. As Withrow advises with chatbot humor, “You’re letting the users make the jokes. Otherwise,” she says, “you’re playing the straight character to their funny.”
It’s this prospect of a shared humorous experience, as opposed to merely setting up the chatbot’s jokes, that seems more aligned with our expectations of human relationships. In “Humor in Romantic Relationships, a Meta-Analysis,” University of Kansas associate professor of communications studies Jeffrey Hall examined the findings of 39 studies to determine the importance of humor in relationships. He determined that shared laughter and humor sensibility was more important than one partner being a jokester. This playfulness was considered crucial in forming bonds.
If our attraction or repulsion with chatbots is similar to what we like or dislike in a romantic relationship, it is illustrative to consider what we expect with another person. Speaking to the university’s publication KU Today, Hall pointed out the important difference between one person joking and two people sharing a laugh. “People say they want a sense of humor in a mate, but that’s a broad concept,” Hall said. “That people think you are funny or you can make a joke out of anything is not strongly related to relationship satisfaction. What is strongly related to relationship satisfaction is the humor that couples create together.”
Withrow thinks a lot about this mutually-created humor. “You can do it with the personality of the chatbot and the user,” she says, pointing to funny choice buttons as a smart feature. Instead of having a chatbot joke as a central feature, a humorous choice button allows the user to decide whether or not they go down a “jokey” path.
When inserting humor into the conversation, Withrow is a fan of using GIFs, and she uses Giphy to find them. “A lot of companies have become wise about using GIFs,” says Withrow, who emphasizes the format’s interdependence between text and visual that is ideal. “GIFs can create an emotional connection and be illustrative.” GIFs also have the benefit of dramatically reducing the need for text, which helps most chatbot developers who may have an impulse to overwrite the copy.
Another major issue with chatbot humor is when to utilize humor. Withrow strongly advises against having humorous responses when the user goes off-script. Most chatbots are based on a flowchart of expected questions and built-in replies (if this, then that, or “IFTTT”), where the chatbot is also trying to keep the conversation within a set of parameters. For example, a chatbot set up to help users find clothes to buy wants the questions about clothes. What happens when users ask the chatbot about its age, hobbies, or whether it wants to go out on a date? Chatbot users commonly go off-script either by accident or during the inevitable testing of the chatbot’s capabilities and parameters. “Keep the humor in the main part of the bot to encourage people to use the main feature,” says Withrow. ”If the user responds off-script, don’t respond off-script.” This also allows the chatbot to maintain a tone, instead of being defined by a rogue user.
There is also an important balance between chatbot personality and utility that could impact user satisfaction and retention. As important as developing a “personality” for the chatbot may be, it could be trumped by the user’s need for utility. While a chatbot creator may want the user to view the chatbot as a friend, the user is still typically trying to accomplish a utilitarian goal. They are using the chatbot for a purpose, not to combat loneliness. That purpose may be finding out a piece of news, sports scores, or an item to purchase. “We use personality and humor sparingly so it doesn’t get in the way of utility,” says Withrow. “If the personality gets in the way of utility, people will abandon ship.”
When Larissa Banting creates chatbots with personality for her company BizzyBots, she favors self-deprecating humor. Banting also shies away from religion and politics. As an example of a widely-used chatbot that she believes strikes an appropriate balance, Banting points to Poncho—a chatbot that delivers the weather with a humorous tone.
In order to strike the appropriate tone, however, the chatbot creator truly needs to understand their users. Who are they? What are their sensibilities? To determine this, Banting sends all of her clients a questionnaire to determine what kind of voice the chatbot should have. After the chatbot creator understands their audience, they can then align the humor.
“People want to buy from people they like,” says Banting, who states that adding humor to a chatbot is humanizing the interaction. “Make it so that the reader is looking forward to it.” Banting mentions the benefits of humor to make the user feel comfortable, and work towards establishing trust. It is this trust that Banting finds most important. “People are resistant to marketing,” she says. “Humor breaks down the barrier.”
Humor certainly breaks down barriers and makes a chatbot appear more human. The trick, however, is making the humorous dialog a conversation that other humans truly want to have.