News flash… the digital space is ever evolving. No place is this more apparent than the onslaught of artificial intelligence implementations that are quickly (and subtly) injecting themselves into our everyday lives. When Siri was introduced on the iPhone back in 2011, AI was looked at as a curiosity. But as these services have become more common (I’m looking at you Alexa and Google Assistant), and our interactions with them more comfortable, businesses are looking at the benefits of these tools in smoothing customer interactions. If you haven’t already done so, now is the time to consider four areas where AI implementations can improve user experience.
Some of the earliest implementations of AI were applied to search, and this makes a lot of sense. Search, even in its simplest incarnations, is an algorithmic exercise. In basic keyword search, for example, when I type the word “dog”, an algorithm stores that string (a collection of characters), evaluates the available dataset for the same string, and returns results that contain the string, in this instance, “dog.” As search has matured over the decades, more and more parameters have been combined with that basic keyword approach to the point where a Google search is evaluating a single word against over 200 parameters.
Search is the most obvious place to use AI to augment your current digital offerings both internally and externally. Brand-name services such as Watson from IBM, offer capabilities that can augment your pre-existing keyword search with additional components such as natural language recognition and semantic relationships that can make entering search criteria simpler and the returned results more relevant.
When Facebook rolled out chatbot functionality to its messaging services a couple years ago, the ground shifted beneath the customer service industry. Although chat-based AI hasn’t completely replaced a responsive human customer service representative, the implementations of these experiences are expanding. Most likely, you’re already interacting with digital virtual assistants if you’re engaging in chat with a major telecom service such as Comcast or AT&T. (In fact, I was just texting with “Samuel” from AT&T this week).
Chatbots can be used in two valuable ways: assisting potential new customers in converting to a purchase, or helping existing customers with problems that may arise when they’re using a good or service. Even the exercise in developing the personality of the bot, the script it will use and the services it will address can help your company get to a better understanding of the questions and concerns your new and existing customers may experience with your products and your brand.
One caveat regarding chat: the successful implementation of chatbots also means being structured in such a way that the assistance the bot provides is backed up by great real-world service. If your bot is appropriately funneling leads to your sales team, it means nothing if those leads aren’t acted upon by a representative from your firm.
Companies small, medium and large are constantly struggling with the proper organization and classification of their content. In fact, errant content tagging is often one of the more common issues we hear about when we’re onboarding new clients. Returning the proper content when called is crucial to delivering a quality user experience.
Content classification issues often fall into three categories: none, all and legacy. “Nones” are companies that have never had a need to segment their content, so it’s not labeled at all. “All’s” are just the opposite – their content is often over-tagged so every piece of content is returned on every piece. “Legacy” companies have content that has been around a while and have migrated systems a couple of times. Their content is tagged, but it’s loaded with outdated tags that are now irrelevant to the new aim of the business.
Regardless of which category your business falls into, there is good news: services exist and are emerging in the marketplace that capitalize on machine learning to help you better organize existing and new pools of content – for video, images and text. Machine learning is exactly what it sounds like: using a small pool of starter criteria, you train a program to accurately read and identify content. The program will get better at identifying the inputs over time and will begin to return better and better results.
Sorting thousands or articles and images is a daunting task for humans that is far better for machines. Companies like Clarifai use neural networks to make it easier for companies to automatically classify and tag custom libraries of imagery and video. AWS offers a variety of tools to recognize, evaluate and categorize a wide array of content, from the spoken word to long-form text.
When the Associated Press announced in early 2015 that they were beginning to use bots to author some categories of articles, it rocked the world of journalism and confirmed the worst nightmares of countless publishers had come true… the robots were here. In the ensuing three years since that announcement, more and more work has been done in the field of content authoring, across a wide array of media—writing, music, and painting are a few examples.
Content authoring is in some ways a more complex endeavor than the other three categories listed on this page, but it only takes a little imagination to see a world in which certain types of media are generated, deployed, measured and reported on by machines, which then evolve those pieces over time to create more and more targeted variants. In fact, our own Justin Daab pointed to the evolving perception of “creative” work last summer.
It’s time to embrace AI
The tech has matured to the point where it’s time to be thinking seriously about incorporating AI in your digital ecosystem. Adding just one or two of the features mentioned above can greatly improve the user experience of the products and services you offer. More importantly, it puts you and your teammates in the mindset of thinking more broadly about what’s possible, and sets you up for a future where these features aren’t simply add-ons… they’re table-stakes.