Siri, Alexa, Cortana, Watson, Zo, Buddy. As AI bot making continues to take off, bot developers and marketers have been keen on giving their bots names that sound human. They’re also giving them a face and their own distinct personality.
However, as intelligent bots or virtual assistants move towards becoming an $11 billion industry by 2024, bots with personalities, which have been programmed from the inside out — rather than the outside in — have been behaving badly.
Many of these problems have derived from a common issue: While bots understand what a human may be telling them at a baseline level, they fail to understand the intention and personality of the human they are conversing with.
From a scientific perspective, some may say that the shortcoming is in natural language processing (NLP) — the science of extracting intention from the text itself. However, I’d argue that they are failing because of the 7 Percent Rule.
The majority of chatbots today are text-based, which means they are unable to analyze the other 93 percent of communication that occurs outside of the A-Zs. They’re simply unable to access that 55 percent of communication that can be analyzed through body language or the 38 percent of context that can be gleaned from the tone and tempo that someone uses.
Understand human interaction with personality buckets
What is a bot to do without access to these non-language-based cues in text-only channels? While current bot designers have been hung up on giving their bots personalities through the five-factor model of human personality, which includes agreeableness, conscientiousness, neuroticism, extraversion, and openness, I’d argue they’d be better off focusing on the other direction: enabling their bots to understand the different types of human personalities they will encounter.
Luckily, there is a human personality model for bot makers to use. In the 1970s, Dr. Taibi Kahler developed the process communication model (PCM) from his work with transactional analysis. NASA adopted this process to assist in astronaut selection in the 1980s and funded the creation of the Personality Pattern Inventory (PPI), which identifies the user’s order of preference for each of the personality styles.
PCM is the only personality model that uses language as an important window into successful communication and interactions, as well as human behavior. Language indicators include keywords, phrases, tone/tempo, grammar, and syntax. By analyzing a person’s communication preferences, PCM identifies their personality as fitting into one of six categories: organizers, connectors, advisors, originals, doers, and dreamers.
The most common personality group in North America is connectors. Connectors make up for 30 percent of the North American population and are hardwired to view the world through a lens of feelings, emotional states, and people. They value relationships and seek unconditional acceptance. Their language reflects their need to not put their relationships at risk.
The second most prevalent group in North America is organizers. About 25 percent of the population is comprised of organizers, who are focused on time structure, organization, and linear problem-solving. These are the individuals at work who place a high priority on to-do lists and gain a sense of accomplishment and success in crossing things off of them.
Another group to highlight is originals. An original is your creative type. They’re playful, fun, energetic, and they think outside of the box. They represent 20 percent of the population.
Customize messages for each personality type
With these personalities in mind, some bot builders may be thinking that they should be aiming to create a bot that communicates best with connectors, given that they are the largest personality group. However, in doing this, you’d likely be alienating the five other personality types with different communication preferences.
These communication preferences are as different as the people who make up the groups. Therefore, a chameleonlike bot that is able to analyze these six different personality types and interact with each accordingly is the best case scenario.
For organizers, it boils down to fixing the problem they have in an efficient manner. In a sales environment, the devil is in the details. So if your bot was attempting to sell a new iPhone to an organizer, a well-received message might look something like this: “It comes with wireless and fast charging … and is preloaded with apps to organize your life … new long-lasting battery minimizes your downtime, so you can keep working.”
The original, on the other hand, likes the newest, coolest, hippest product out there, and they want it to be easy to use. Therefore, a well-received bot message to them would look something like this: “Wow! … chat, selfie, share, play, and whatever else you want to do when you want to do it! It’s all set up for you.”
Meanwhile, a connector is looking for a friendly bot that conveys a personalized message of how the new iPhone will let them connect better with their friends and family. For them, a message like this would serve: “Hello again, it’s so good to hear from you. You’re going to love this … the new iPhone keeps your family connected.”
Evolve empathetic bots
While these examples may seem somewhat rudimentary, they illustrate a first step toward training human personality analysis into your NLP. Once this is fully baked into your chatbot, it can serve as the core engine driving a more empathetic bot. A bot built to respond based on the personality and intention of the human on the other end of the conversation.
After all, the real barometer for a chatbot has not changed much since the first chatbot — SmarterChild — debuted on AOL’s Instant Messenger in the early 2000s. At a basic level, all bots are designed to simulate human conversation.
Therefore, the best user experiences occur with natural and authentic dialogue between bots and humans. By adopting baseline personalities for humans, bot makers can design AI so that the dialogue is more natural and customized.
As we look to the future, where chatbot conversations will take on audio and visual forms, our ability to accurately analyze human personality by leveraging cues outside of language will get that much better.
In this future, the bots that thrive won’t be the ones that were designed to have unforgettable personalities, but rather those that truly understand the personalities they are communicating with.
Christopher Danson is chief technology officer at Mattersight, a company that provides SaaS-based enterprise behavioral analytics software.
VentureBeatVentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact. Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:
- up-to-date information on the subjects of interest to you
- our newsletters
- gated thought-leader content and discounted access to our prized events, such as Transform
- networking features, and more