Every AI system has a little bit of personality, even if it doesn’t always seem that way. How systems are equipped to handle conversations with humans depends largely on how they’re trained and whom they’re built to serve. Read this post to learn more about how to deliver the proper personality and psychology for your AI system.
Any Artificial Intelligence (AI) deployment should begin with a simple question: Who is going to interact with the technology? Starting from there, you can move onto more business-focused questions: For which use case should we build? What problem are we trying to solve? How will this impact revenue? The psychology of your AI experience will depend almost entirely on your intended audiences, what you hope for them to achieve, and how this experience will redefine your business.
We’ve written in the past about designing an AI system for teenage gamers and why that experience should be dramatically different than an in-house system used primarily by insurance agents. Building a system for gamers allows for creativity, fun and humor, whereas business users want immediate results without any considerable barriers from within their AI experiences. Building a fun, colorful and humorous system for business users could work, but you risk alienating customers with serious and timely concerns. Conversely, a results-oriented AI experience for gamers could be boring, or off-brand, which could turn off users to the technology.
The psychology and aesthetics of an optimal AI system deployment will be one of the many topics discussed at this year’s Digital Workforce Summit (DWS) on May 8th in New York City. We’re bringing together experts from around the world to focus on the how, when and why of AI systems — and the answers are not the same for every business. In this post, we’ll explore some of the other areas you should consider when determining the psychology and style of your AI build.
Specific Language and Personalities
How you’re treated when you walk into a high-end jewelry store should be different than when you walk into a fast food restaurant. Both businesses should treat you with respect, show enthusiasm to provide service and work hard to ensure you get what you need. However, jewelers should work with you to build very specific items that are catered and crafted to your every requirement. They should ask you questions about how you’d like your jewelry designed, how you’d like it to fit, how much you’re willing to spend, etc. If a fast food server provided the same level of service to each customer, the restaurant would have to cease referring to itself as “fast” food.
The language AI systems use to interact with customers should vary from project to project. You can program your AI system to use humor, to be personal and engaging, or you can program it to speak in direct terms, e.g., “Would you like to add fries and a drink to that for an additional dollar?” Whereas, if your system is a hub where people interact with products and services, and you want them to spend time within the environment exploring options and services, you’ll want a more affable AI-based personality.
Location and Channel Matters for AI Deployments
Where your customers or users interact with your system is almost as important as how your system addresses them. Is your AI system a chat-based platform that’s built to live on your website? Do you intend for it to be an overlay for your mobile application? Do you want to take your AI directly to customers within messaging apps like Facebook Messenger? Maybe you’ll instead focus on the voice experience and build your AI system for phone calls or for commands on voice assistants like Alexa.
There is a crucial reason that this needs to be determined prior to AI deployment. It has to do with the data that’s available to the system, how the data is used, what the system is able to accomplish and the way the system responds. For example: If you want to build an AI system to help elderly patients diagnose medical issues, you can’t just plop your AI into Facebook Messenger. Firstly, the data being passed from user to company needs to be secured in a way that Facebook or any social media platform may not be able to guarantee. Secondly, Facebook might not be the ideal location for elderly users to interact with your technology for obvious reasons. Instead, building your experience for telephone-based communication may be the ideal set-up. Most elderly patients are more familiar with how to use a phone than they are social media, and you can guarantee security and privacy.
Conversely, building a phone-based AI system probably will not cut it for teenage gamers. Kids don’t talk on the phone as often as previous generations. Additionally, they’ll want to be able to share visuals from their gaming experience with the AI system to detail any issues, or to share experiences they would like to recreate. None of this is possible via a phone call. Also, most kids don’t expect or necessarily care about a good bedside manner; they want fun and information, and they want it delivered within the confines or their natural gaming or browsing experience.
The psychology behind your AI system largely depends on the psychology of the person making use of the technology. Aligning your users’ expectations with how the AI system looks, speaks and serves will benefit them and your company. Deviating in tone, style, or substance from the user’s expectations in turn could prove costly.
You can hear about how global enterprises tackled these and other issues with their AI deployments at this year’s DWS.