Google invents a discipline to help us make perfect decisions

[jrineakter]

Cassie Kozyrkov is Google's Chief Decision Officer, a key position within the tech company. Kozyrkov has trained 17,000 Googlers in decision making by making data more efficient through psychology, neuroscience, economics and business management. And now Google wants to share this new discipline, Decision Intelligence Engineering , with the rest of the world.

Kozyrkov's figure became indispensable at Google after the California company adopted machine learning technology in some of its products and services, a learning model that decides –for example– if a photo of an animal is a cat and then acts accordingly: if it is a cat, A is done. If it is not a cat, B is done. These are automated decisions without the intervention of a human.

The problem with this machine learning is that the algorithm is only as good as the trainer who puts it to work. That is, if the human training the algorithm shows it a picture of a dog and tells it that it is a cat, then it will perform incorrectly: the more sophisticated the machine learning, the more opportunities there are for humans to introduce subtle problems into the final results.

So Google needed a decision-making venezuela number data framework that would enable humans and machines to make informed decisions based on accumulated data. Such a framework didn't exist, so they set out to build one.

The science of a good decision
The academic field of decision science encompasses psychology, neuroscience and economics, but it does not touch on the engineering and scaling of automated decision making, just as data science is not directly linked to human thought and emotional impulses.

“A big part of the training that data scientists go through is that decision makers know exactly what they need, ask the right question, and frame the problem perfectly,” Kozyrkov explains, as reported by Fast Company . “A data scientist collects information to put it to work on that question: answer it or build a machine learning system to solve it.”

But the scenario of perfect decisions, in which the right questions are always asked, is completely unrealistic. While working with the data science division at Google, Kozyrkov often saw executives making decisions that were driven by unconsciousness rather than by accurate, rational data.

From there, Google launched a graduate training program that, rather than simply training decision makers in data science, set out to draw on behavioral science to make truly data-driven decisions. This means being able to identify a bias and get around it, to frame a decision effectively, often before even looking at any data.

How do you make an effective decision?
The first thing Google asks decision makers to do is to determine how they would make the decision without any additional information. What would be the intuitive choice? You have to decide whether to dine at a restaurant. You've seen the dishes but no one has told you about it, either good or bad; based on the photos, do you stay for dinner there?

“We often think that we don’t have a preference, but we’re actually lying to ourselves,” says Kozyrkov, referring to the cognitive bias that we all camouflage in apparent neutrality. “We have some kind of intuition about what seems to be the safest option in situations of uncertainty.”