I get this question almost daily. Does it sound familiar? I’m referring to when the advertising industry started to implement programmatic solutions, when there were multiple definitions for the word, “programmatic” and even more flavors of technologies claiming the category.
The word “programmatic” means different things to different people, but I go by the following definition:
Programmatic Advertising: The automation of workflow and data-driven decisioning, which creates customization of inventory or audiences purchased to receive specific messages. This may or may not incorporate real-time-bidding (RTB). A programmatic solution also may or may not include machine learning or AI capabilities.
AI is a term that is bandied about without strict adherence to one meaning. Machine learning is also a term that is used loosely, sometimes referring to AI -- but the distinction between these technologies is worth understanding.
Machine learning generally relates to a software algorithm that leverages specific data sets and rules programmed by humans that become faster and smarter as they iterate in performing the tasks for which they were created. There are numerous ad-tech and martech solutions today that use some form of machine learning.
Artificial intelligence (AI) exists when software can make decisions outside of the data strategy and rules set by humans, to form new solutions. AI is nascent in the marketing technology space, and there are far fewer companies with developed AI technology.
Frank Speiser, co-founder and chief product officer of Socialflow, shared a useful analogy that separates AI from machine learning: If you are a farmer using a machine learning tool, you might use it to help optimize the planting of your crops. It could tell you when you should plant your corn and tomatoes, and when to water them to get the best result. Over time, the advice given becomes more and more accurate as the algorithm gets smarter about weather conditions and the aggregate yields of corn and tomatoes.
AI would go beyond this capability, possibly suggesting that you plant a different crop altogether, or by telling you that planting crops is the wrong activity given the success of sheep farming in the area.
The Socialflow algorithm offers some similar “thinking.” For instance, the machine learning aspect of the platform will bucket audiences who respond to specific content in behavioral categories defined by the developers, for example, travel, automotive, cooking, entertainment, etc.
But when enough responders don’t fit into a predefined category, the tool will create a new one — say, “Shoe Aficionados” — and will then alert the developers about the new segment to consider. In this case, the algorithm isn’t just doing the job it was told to do by humans, but rather doing the job of the humans in refining the data decisions that can/should be considered.
There are many aspects of AI that separate it from other kinds of rapid data-decisioning technologies such as image and voice recognition, and contextual understanding of topics: non-numerical things that the human brain finds really easy, but where computers typically struggle. Check out Pawel Sysiak’s essay, “AI Revolution 101,” a great short read if you want to fully understand the three tiers of AI:
ANI: Artificial Narrow Intelligence
1st intelligence caliber. “AI that specializes in one area. There’s AI that can beat the world chess champion in chess, but that’s the only thing it does.”
AGI: Artificial General Intelligence
2nd intelligence caliber. AI that reaches and then passes the intelligence level of a human, meaning it has the ability to “reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly, and learn from experience.”
ASI: Artificial Super Intelligence
3rd intelligence caliber. AI that achieves a level of intelligence smarter than all of humanity combined—“ranging from just a little smarter … to one trillion times smarter.”
When you are inevitably asked, “What do you mean when you say AI?” hopefully now you will have a better answer.