Esprii Chapman – Notes on Prediction Machines – 12/12/20 (Book 5 of 5)

a prediction is not a decision. Making a decision requires applying judgment to a prediction and then acting. Before recent advances in machine intelligence, this distinction was only of academic interest because humans always performed prediction and judgment together. Now, advances in machine prediction mean that we have to examine the anatomy of a decision.

Agrawal, Ajay. Prediction Machines (p. 81). Harvard Business Review Press. Kindle Edition. 

decisions have six other key elements (see figure 7-1). When someone (or something) makes a decision, they take input data from the world that enables a prediction. That prediction is possible because training occurred about relationships between different types of data and which data is most closely associated with a situation. Combining the prediction with judgment on what matters, the decision maker can then choose an action. The action leads to an outcome (which has an associated reward or payoff). The outcome is a consequence of the decision. It is needed to provide a complete picture. The outcome may also provide feedback to help improve the next prediction.

Agrawal, Ajay. Prediction Machines (p. 81). Harvard Business Review Press. Kindle Edition. 

Agrawal, Ajay. Prediction Machines (p. 82). Harvard Business Review Press. Kindle Edition. 

Imagine you have a pain in your leg and go to the doctor. The doctor sees you, takes an X-ray and a blood test and asks you a few questions, resulting in input data. Using that input, and based on years in medical school and many other patients who are more or less like you (that’s training and feedback), the doctor makes a prediction: “You most likely have muscle cramps, although there is a small chance you have a blood clot.”

Alongside this assessment is judgment. The doctor’s judgment takes into account other data (including intuition and experience). Suppose that, if it is a muscle cramp, then the treatment is rest. If a blood clot, then the treatment is a drug with no long-term side effects, but it causes mild discomfort for many people. If the doctor mistakenly treats your muscle cramp with the blood clot treatment, then you are uncomfortable for a short time. If the doctor mistakenly treats the blood clot with rest, then there is a chance of serious complications or even death. Judgment involves determining the relative payoff associated with each possible outcome, including those associated with “correct” decisions as well as those associated with mistakes (in this case, the payoffs associated with healing, mild discomfort, and serious complications). Determining the payoffs for all possible outcomes is a necessary step for deciding when to choose the drug treatment, opting for the mild discomfort and reducing the risk of a serious complication, versus when to choose rest. So, applying judgment to the prediction, the doctor makes a decision, perhaps, given your age and risk preferences, that you should have the treatment for the muscle cramp, even though there is some tiny likelihood you have a blood clot. Finally is the action in administering the treatment and observing the outcome: Did the pain in your leg go away? Did other complications arise? The doctor can use this observed outcome as feedback to inform the next prediction.

Agrawal, Ajay. Prediction Machines (p. 82-83). Harvard Business Review Press. Kindle Edition. 

As machine prediction increasingly replaces the predictions that humans make, the value of human prediction will decline. But a key point is that, while prediction is a key component of any decision, it is not the only component. The other elements of a decision—judgment, data, and action—remain, for now, firmly in the realm of humans. They are complements to prediction, meaning they increase in value as prediction becomes cheap.

Agrawal, Ajay. Prediction Machines (p. 83). Harvard Business Review Press. Kindle Edition. 

With better prediction come more opportunities to consider the rewards of various actions—in other words, more opportunities for judgment. And that means that better, faster, and cheaper prediction will give us more decisions to make.

Agrawal, Ajay. Prediction Machines (p. 91). Harvard Business Review Press. Kindle Edition. 

  • They’re not to the point of being able to predict and judge yet.

There wasn’t much that was really super good from this book.

Agrawal, Ajay. Prediction Machines. Brighton, MA, Harvard Business Review Press, 2018. 

1 Comment

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s