#marketing automation

If every single marketing automation software or “one more” photo filter app on iTunes comes tagged with AI.

True, with neural networks we can teach computers to classify information, in the same way, we “as humans” do —but generalised AI examples, are few and far between.

What we see today is applied AI, where one particular subset of it, machine learning, is making rapid progress. The machine “learns” based on large sets of structured data which map out various cause and effect scenarios.

Think of it very simply as a scaled up version of the hugely popular IFTT app.

IF THIS, THEN THAT, based on a system of probability and executed at scale.

Most of martech, programmatic ad-tech and bots – work on the above principle of machine learning. Large sets of structured data get pared (cause and effect) in a system of probability.  Execution happens when a specific activity triggers the system to drive automated workflows based on the above probability model.

The difference and hence benefit here is the sheer scale of things. Humanly one can automate a handful of workflows based on cause and effect scenarios. In machine learning, this gets executed at speed across multiple permutations and combinations of probability at the same time.

What we need to note is that the system still executes commands based on what it has been “taught”. It does not have the capacity for independent thinking - yet.

Slapping AI across marketing messages to sell automation software is probably a bit misleading.

Most photo apps don’t work on AI in its real sense.

Neither do most chatbots.

Nor do website builders.