There is nothing wrong inherently with relying on Google's and Facebook's targeting technology. They have really good tech - the best ever made in some cases. Most of the time the work of a platform in interfacing with Google and FB is in delivering the right data to them and then setting up guardrails to allow for industry-specific needs that will deviate from what their algorithms say (i.e. a sale of X model may be more valuable for reasons like stair step, OEM preference, etc.)
@ryan.leslie is right that encoded knowledge is both non-sexy and technically defined as AI.
Instead of dunking on the non-sexy forms of AI, though, it makes more sense to ask about the applications in which it’s being applied.
Let’s say I’m using AI to decide what offer to show on a dealer’s website to what visitor. I could use encoded knowledge/IFTTT and say IF visitor looked at page A, THEN show them offer B. That’s hard-coded and I could probably get pretty good results if I worked on getting as close as possible to the right formula.
If I used machine learning, I could get MUCH closer. That’s saying ‘lets look at the million previous interactions and outcomes and all their various attributes, and then based on those results decide what to show the million + 1 visitor’. That’s pretty unbeatable.
The question to ask is, what’s the incremental value of ML above encoded knowledge? For some applications IFTTT can suffice, and that doesn’t mean it’s not valuable.
Finally, obviously using automation doesn’t make sense in all situations—there are many places where human intervention is best. But there are many many places where automated products, with really good design, can do much better than what humans are capable of, both in terms of cognition and scale.
I've seen an executive make AI a huge initiative for (unnamed company) 12-18 months before it could actually be a part of their platform, constantly preach about how they're adding AI to their platform, have employees (who have nothing to do with product or engineering) take ML/AI training so they can say they know what AI is, then create HUGE problems for product and engineering teams by demanding they focus on adding AI to the platform instead of focusing on stuff that actually matters and things that dealers, the sales team, support teams have all been asking be added to the platform for several months if not years and totally blow up relationships with tons of customers by ignoring enhancements that aren't AI. It's asinine.