Repeat after me: Machine learning is not new.
It isn't. In fact, Arthur Samuel, the giant in computer gaming and artificial intelligence, first coined the term “machine learning” in 1959 while he was with IBM. It started as pattern recognition and now represents the ability to develop algorithms that learn from and can make predictions based on data. For the truly nerdy among us, there is an argument that machine learning is synonymous, or more likely overlapping, with computational statistics.
But none of this is new.
It just feels new because instead of being confined to the comforts of the IT team, machine learning has evolved into a vital tool for customer engagement and experience. Machine learning has been unleashed on the marketer. Now…if we could just figure out how it is different from artificial intelligence and cognitive computing.
Thanks to a recent CMO Council webcast, I have some details to share to shed a little light on what I’m talking about:
- Machine learning is a branch of artificial intelligence in which a computer generates rules underlying or based on raw data that has been fed into it.
- The process of learning begins with data as the system looks for patterns in order to make better decisions in the future.
- Machine learning’s primary aim is to allow the computer to learn automatically without human intervention and then create adjusted automated actions accordingly.
- Machine learning enables analysis of massive quantities of data.
- But…machine learning is not the same as artificial intelligence (AI) or cognitive computing.
- Artificial Intelligence is a system featuring human-like intelligence in the form of an autonomous technological system. AI is most often defined as an "intelligent agent"—specifically any device that perceives its environment and takes actions that maximize its chance of success at a particular goal.
- Cognitive computing is most recently defined as a computing system that mimics the functioning of the human brain (although I’m also going to share the best ever entry in Wikipedia for a marketer: “At present, there is no widely agreed upon definition for cognitive computing in either academia or industry . It is most frequently used as marketing jargon.”) #ouch
- Most often, cognitive computing displaces specific features that are adaptive, interactive, iterative and contextual, separating its learning capabilities from machine learning.
So how does it all come together? Here is the real brain-twister: Cognitive systems can use machine learning to adapt to different contexts with minimal human supervision. That’s right…while machine learning isn’t cognitive, cognitive leverages machine learning in order to ingest and analyze data so that it can learn and adapt to context.
For the marketer, cognitive can mean improved insights and engagement with customers, or it can further punctuate the fact that analytics goals to date are far off, making it feel difficult to advance an initiative like cognitive. When we asked marketers about their initial reactions to cognitive, 28 percent said they were excited about the possibilities, from improved targeting and segmentation to real personalization and engagement. An additional 16 percent are feeling a mix of emotions ranging from excitement to relief at the prospect of finally making sense of the data in their systems.
However, it is telling that 24 percent are not quite sure what cognitive computing is.
If you are part of this 24 percent, I would recommend listening to our on-demand webinar featuring Brady Fox from IBM’s Watson group, who shares some thoughts, definitions and examples about cognitive computing. I stole—nay, “borrowed”—much of the technical detail in this note from his presentation. If you missed the webcast, check out the on-demand version here.
We will keep having conversations on everything from AI to cognitive computing as data and the demands of the customer experience evolve.
Until next month!