Insights & Tips

Already a subscriber? Login

Become a subscriber and unlock an information arsenal focused on making your pricing efforts more effective.

The One Thing Pricing AI Still Can’t Do

When I was attending college, I heard a story about a group of guys who thought they had a ‘cheat code’ for one of those Bible-as-literature classes that were popular back then. Before signing up for the course, these five roommates had learned that the final, blue-book essay question for the last three cohorts had been, “What’s the significance of the Sermon on the Mount?” And knowing this, these roomies assumed they could just focus their preparations on that one question and rack up some very easy credits.

Of course, when it came time for the final essay test, the professor wrote this question on the chalkboard, “What’s the significance of kicking the merchants out of the temple?” Recognizing that their scheme had been foiled and they weren’t prepared for this question, four of the roommates just got up and left the classroom in utter defeat.

The fifth roommate, however, was the only one to get a passing grade. Not a great grade, mind you, but a passing grade. How? Well, rather than give up altogether, he started his essay by writing, “While kicking the merchants out of the temple was an important element of the story, I would argue that its significance pales in comparison to the Sermon on the Mount.” He then proceeded to write his essay on the significance of the Sermon on the Mount, as originally planned!

I was recently reminded of this story by a subscriber’s questions about artificial intelligence. The subscriber wanted to know whether an artificial intelligence engine would do all the right things and provide all the right answers without being told or asked. It’s certainly not an unreasonable question, given the extreme levels of puffery and hype we’ve been seeing in the media around all things “AI.”

But here’s the thing…I’m not aware of any current AI application that will do what the fifth roommate in the earlier story did, and effectively say, “Hey, this stuff you want me to do and this information you want me to provide is kinda nice and all; but it’s relatively weak and unimportant compared to all this other stuff you aren’t doing and aren’t asking about.”

To be clear, artificial intelligence is an incredibly powerful enabler that can help automate a lot of extremely complex processes and answer a broad range of very complex and weighty questions. And it can do all of this much faster and more accurately than even the largest teams of human savants imaginable.

That being said, as some of the recent high-profile “whoopsies” have made abundantly clear, even the most sophisticated artificial intelligence applications still require a great deal of expert guidance and thoughtful direction. Behind the scenes, smart people still need to configure the models and provide the parameters to tell HAL what to do, what not to do, where to focus, what to prioritize, and so on.

Simply put, artificial intelligence can answer almost any question you might ask. And AI can execute almost any task you might define. But like it or not, it’s still up to you to ask the most valuable questions and define the most valuable tasks!

That’s why specialized, AI-enabled pricing applications will continue to be the weapon of choice for leading teams, even as artificial intelligence engines, models, and libraries become more and more ubiquitous. You see, instead of having to know all the most valuable questions to ask and all the best tasks to define, pricing teams can leverage the knowledge, expertise, and experience that has been aggregated, tested, refined, codified, and productized by the specialized pricing technology vendors.

Eventually, we’ll get to a place where AI can do all the right things and provide all the right answers without ever being told or asked. But we’re just not there yet.

PS: Along the way, I know that some readers will have muttered to themselves, “Yeah, but even if we don’t know, we can just ask the AI to list out all the most valuable questions and tasks.” And without getting into the weeds on the training dataset that would be required, I’ll just highlight how asking the question proves my point.

Get Immediate Access To Everything In The PricingBrew Journal

Related Resources