Tech Minute: Machine Learning, Bots, and Cognitive Web Services – Oh My!

January 23, 2017

Share

Could they be the “great and powerful Oz” for brand marketers?

It was recently announced that Estée Lauder is the first major retailer to launch a Facebook Messenger bot service. Using artificial intelligence, the bot provides holiday shoppers with gift ideas and a 60-minute delivery service all from the comfort of Messenger. Many similar bots have been announced by brands in the last year. As a developer working for a marketing firm, this got me thinking. If we can go to Facebook Messenger for immediate Christmas gift delivery, what other brand marketing opportunities are now within reach? What’s up with all these bots? Is this the rise of SkyNet that those Terminator movies were warning us about? (No!)

Over the past year, Cloud service providers like IBM, Google, Amazon, Apple and Microsoft have been announcing new machine learning (ML) or artificial intelligence (AI) technologies.   We are all becoming familiar with the specialty Agent Bots like Siri, Cortana, Alexa whose ‘skills’ are based on connections to online services via an API. Most recently, the new Amazon Go checkout service will allow its grocery store shoppers a no-waiting checkout experience through its use of ML, AI, and sensors.

Some of these ML technologies are being offered as cognitive services and are available for developers and marketers to very easily teach and personalize for the specific needs of a brand. Cognitive Web service APIs are growing in popularity. Microsoft’s Azure platform now offers numerous cognitive service APIs around language, sentiment, facial recognition, emotion, vison, search and more.   For example, media sentiment through ML can now be defined within the context of what the specific brand teaches it. Thus making sentiment unique for a given brand based on what the machine is taught. This kind of teaching can increase the accuracy of sentiment analysis and account for brand specifics around context, sarcasm, and slang.

Now here’s where things get interesting for developers and marketers alike. Imagine the capabilities of attaching a chat bot into a cognitive service, customized for a specific brand campaign.   Instead of a defined response from the bot, a response is given from the brand based on a consumer’s facial expression, speech verification, uploaded image, video background, language, or location.   Brand bots could even be modeled to determine ‘intent’ through the use of language understanding cognitive services. The possibilities become almost limitless for immediate customer service and customer relations. Taking b-to-c communication to the AI level.

For example, let’s say Dorothy’s Ice Cream want to start a campaign to determine the best new ice cream flavors.   The campaign goal is get as many images and videos of people eating their new ice cream flavors to determine how the new flavors make the custom feel. A chat bot is developed with facial recognition and emotion detection cognitive services. As users upload their videos and images, the bot immediately detects the type of emotion experienced from each flavor of ice cream and respond to the user with additional questions or promotions. The emotion of ‘disgust’ obviously wont’ warrant that half-off coupon.

From a customer service perspective, bots could assist overwhelmed phone support during times of high volumes.   “Your estimated wait time is 15 minutes; would you like to try our automated bot to rebook your travel? Go to our Facebook site at…”   The Facebook messenger bot is then modeled and trained to accept travel bookings through language understanding service and send immediate verification.

We are on the verge of something big with bots. Chat bots are here, they are relatively easy to develop, and with machine learning and cognitive services attached, they are becoming smart…Pay no attention to that developer behind the curtain.