There is a lot of discussion about artificial intelligence and machine learning, but should content marketers care? We define the differences and outline the impact for marketers today and in the near future. AI and machine learning are very real but not all that easy to apply effectively....for now.
Artificial Intelligence (AI), Machine Learning, and more recently, Cognitive Computing are getting more and more buzz. From new capabilities on our smartphones (Google Now and Siri) to new devices to talk to at home (Amazon Echo) to computers winning at Jeopardy (IBM Watson), apparently the era of computer intelligence and the end of humanity(Skynet!) is now upon us.
The buzz is about the massive potential these technologies have, but at the same time, the buzz can cause confusion and seriously overpromise on what can really be done with the technology right now. Marketing potential ranges from creating much more relevant personalized experiences for customers, to making better marketing investments in content creation and campaigns, and even ways to make the brand come to life (virtually) as a persona customers can actually interact with.
Artificial intelligence needs application to be truly considered intelligent
Artificial intelligence is a general term for trying to enable computers to make decisions in context. Let’s admit it: right now, computers are crazy useful but pretty darn dumb. I mean, if you leave your computer for a day and come back, what did it accomplish? Maybe download some emails and a security patch? The promise of AI is that technology become easier to interact with; as the computer begins to gain context, it becomes less rigid in how it needs to be instructed and more intelligent in determining what is important—like making sense of those emails, perhaps automatically replying to some and gathering the data you need to reply to others
Learning is key to effective marketing and can be done by a machine
Machine learning is a general term for an approach for programming computers. Instead of telling the computer what to do and how to do it, we tell the computer what we are looking for and give it a bunch of example data so it can use trial-and-error and correlation to identify what we are looking for on its own, getting better over time. For example, if I want to find all the pictures of cats and dogs in my photo library, I would traditionally have to come up with a bunch of rules on how to identify cats and dogs in a photo, which is really hard and error prone, and would never get any better without more coding. With machine learning, instead of coming up with the rules, I just feed in a bunch of photos and mark a bunch of dogs and cats manually, and the computer will learn how to recognize them. For marketing this can be something like figuring out the right product mix to show a new customer, or determine which offer a customer is most likely to respond to. The computer uses a starting point to figuring out the rules on its own, and with a bit of human oversight to correct it where it goes astray, it gets better and better at it. Imagine how powerful this might be for marketers. We can suddenly detect and test things on scale that today would require a room full of interns
The advantages of the machine learning approach are that the accuracy improves with more data, and scaling to additional attributes is much easier. For example, if I want to start identifying elephants in my pictures, traditional approaches would send me back to the drawing board to create rules about how to define elephants. With machine learning, I just need to enter some training data (pictures of elephants), and the computer will learn how to identify them on its own. So it has a single brain that learns from the experience, a benefit that would be difficult to replicate when the experience is spread across many human brains.
IBM and other technology leaders are progressing in the use of these capabilities
Cognitive computing is an category of AI that aims to emulate the human thought process in a computer model (likely machine learning). IBM has been pushing the term “cognitive computing” in relation to its efforts around Watson, so it’s starting to feel a little bit like a branded term more than the specific area of computer research. Marketers are using IBM services to understand personality traits about customers based on limited customer input to make more effective content choices.
We are just at the beginning and advancing rapidly
Technology has been rapidly progressing. Real commercial applications are emerging, and the barriers to using machine learning are breaking down. Industry giants like Google, IBM, Microsoft and Amazon are giving the world access to infrastructure designed to run machine learning. That said, it’s still a very complex process to setup and manage. At this point, don't expect to be able to implement machine learning without a fair bit of programming and technical perseverance.
There are machine learning models already trained and ready to use which can be very helpful to marketers, such as offerings around tone and personality identification from IBM Watson, Natural Language Processing from Google and image identification from Microsoft. These offerings make things much simpler to jump into but still require programming skills to utilize and are very clear in their purpose—which often requires some vision on how to derive business value, and to integrate it into a bigger customer experience.
It works but its not easy
Because of this complexity and the amount of work required to tap into the technologies, there's a lot of disappointment brewing for those that would love to take advantage of their power. Results are not out of reach but do take significant effort to achieve. The complexity will come down in time, and there are the packaged service offerings to make it easier, as well as companies leveraging these technologies to provide value in specific problem domains.
From a marketer’s perspective: there are a lot of solutions out there using machine learning today. Of course, 8-Point Arc uses it for a range of things from identifying emotional tone and content topics to getting image descriptions. IBM's Think Marketing campaign leverages Watson to match content selection with the user's location and behavior to drive relevance, and TD Bank is trying to leverage AI to help understand you as a person, using that context to provide information on financial investing relevant to your knowledge and risk tolerance.