AI’s star grows brighter in marketing, with new announcements from Amazon, Facebook, and Emarsys
Amazon adds a new intelligence-as-a-service, Facebook builds AI for filtering videos, and marketing platform Emarsys is making AI a member of its family.
Several new announcements point to AI becoming as much a standard feature in marketing tools as the cloud, helping to move the age of cognitive marketing closer.
Amazon, of course, has a long history of offering automated intelligence, such as its individualized product recommendations and, more recently, its increasingly popular Alexa intelligent agent.
“We have thousands of people dedicated to AI in our business,” Amazon Web Services CEO Andy Jassy told TechCrunch, validating the importance that the retail giant has assigned to the technology.
Three new services are offered. There’s an image recognition service called Rekognition, which, like other such services, can not only identify visual elements but can also determine their variety, like a dog’s breed.
There’s a text-to-speech service called Polly with 47 male and female voices. Anyone using Amazon’s Alexa voice-based intelligent agent can attest that the company knows how to do a lifelike generated voice.
And, speaking of Alexa, the third new service is called Lex, so that developers can build their own Alexa conversational applications.
The Amazon announcement, retail intelligence platform Rubikloud CEO Kerry Liu emailed me, should propel “a sense of urgency across retailers to incorporate machine learning and AI into their platform in order to meet consumer demands for a tailored experience.”
Vienna-based marketing platform Emarsys has gotten that message. The company, whose US office is in Indianapolis, announced this week it has integrated a layer of intelligence into its platform, now relaunched as Emarsys AI Marketing or AIM. Its 1,500 customers across 140 countries include American Express, Canon, eBay, L’Oréal, Nike and Toys “R” Us.
Emarsys’ new AI layer adds three specific features to the platform, each focused on the individual: recommendations about the most appropriate incentive (like $5 or $25 off); product recommendations; and an automated selection of the best times to send incentives or other marketing material. Here’s what the product recommendations might look like:
CMO Allen Nance told me that the AI enables a marketer to upload and metatag incentives like sales offers, drop a variable placeholder into an email template, and define a strategy. Then the AI determines the most appropriate incentive for each targeted customer, replaces the variable placeholder, adds the appropriate wording to the subject line and sends it out.
The company says this addition of intelligence can increase revenue by 59 percent per thousand customers.
The AI adds predictive modeling to forecast future behavior, Nance said, so it can assess which new products you might buy, instead of just assuming you’ll buy the same ol’ things that your purchase history indicates. It also allows a marketer to automatically run A/B tests that pit AI-driven campaigns against ones without AI.
But, while AI has gotten pretty good at predicting which products or offers you might like, it’s still struggling with determining the harder problem of which content is fake or offensive. And, until the problem of deceptive content can be handled at scale, the credibility of all online content is threatened.
So it was interesting to find out this week that Facebook, which has become a virtual Utopia of fake and offensive content, is working on AI that can filter video material.
Yesterday, the company’s director of applied machine learning, Joaquin Candela, told Reuters that the social networking giant is developing AI that can find offensive video material, such as “nudity, violence, or any of the things that are not according to our policies.”
It’s still in the research phase, he said, as the team struggles with the non-trivial problem of a very fast computer vision algorithm that can understand video as a human does.
But, if AI-based computer vision can accomplish that goal, it could be a huge boon for marketers trying to keep their brands away from the bad stuff — or closer to the good stuff.
Opinions expressed in this article are those of the guest author and not necessarily MarTech Today. Staff authors are listed here.