Power and Prediction: The Disruptive Economics of Artificial Intelligence
Ajay Agrawal, Joshua Gans, and Ami Goldfarb
Harvard Business Review Press (November 2022)

How and why AI requires a focus on system solutions rather than on point solutions

In a previous book, Prediction Machines, Ajay Agrawal, Joshua Gans, and Ami Goldfarb examine the simple economics of artificial intelligence. “AI is a prediction technology, predictions are inputs to decision making, and economics provides a perfect framework for understanding the trade-offs underlying any decision. So, by dint of luck and some design, we found ourselves at the right place at the right time to form a bridge between the technologist and the business practitioner. The result is this book.”

Over time, technologies change but economics do not. The framework provided in Prediction Machines “only told part of the story, the [begin italics] point solutions [end italics] part. In the years since, we discovered that another part of the AI story had yet to be told, the  [begin italics] systems [end italics] part. We tell that story here. How did we miss it in the first place? We wind the tape back to 2017, when we were writing, to explain.” And so they do with meticulous precision. They are superb raconteurs.

In Power and Prediction, Agrawal, Gans, and Goldfarb shift their focus “from exploring neural networks tgo exploring human cognition (how we make decisions), social behavior (why people in some industries are keen to embrace AI quickly while others are resistant), production systems (how some decisions depend on others), and industry structures (how we’ve hidden certain decisions to shield ourselves from uncertainty).

These are among the passages of greatest interest and value to me, also listed to suggest the scope of Agrawal, Gans, and Goldfarb’s coverage:

o Neural networks (Pages xiii-xiv and 10-110)
o Verafin (x-xii 30-32)
o Disruption (10-11, 18-20, 104-105, 107-117, 110-111, 111-1114, and 178-179)
o Systems solutions (13-24)
o Bias (22-23 and 225-235)

o Causation and correlation (27-30 and 36-37)
o Economic issues (45-47, 53-61, 63-64, 77-81, and 156-157)
o Covid-19 pandemic (48-49, 75-81, 78-81, and 97-98)
o Uncertainty (49-52,53-61, 58-59, 63-64, 77-81, and 157-159)
o Greenhouse Systems (59-61)

o Automation (85-90, 121-128, 162-163, and 204-205)
o AlpaFold (97-98 and 101-103)
o Competitive performance (111-114, 119-120, and 131-132)
o First-mover advantages (129-140)
o Decoupling disruption from prediction (143-153)

o Probabilistic thinking (155-166)
o Judgment (160-164 and 167-180)
o Coordination (184-190 and 193-194)
o Modularity (187-190)
o Epilogue: AI Bias and Systems (225-234)

These are among Ajay Agrawal, Joshua Gans, and Ami Goldfarb’s concluding thoughts:

“Today, the individuals who most resist adopting AI systems are those who are most concerned about discrimination. We anticipate that will exactly reverse. Once people realize that discrimination is easier to detect and fix in AI systems than in humans, the greatest resistance to adopting AI will come not from those who want to reduce discrimination but rather from those who benefit most from it.”

While you are reading Power and Prediction, I offer two suggestions for your consideration: Highlight key passages, and, record your comments, questions, action steps (preferably with deadlines), and page references as well as your responses to the questions posed and to lessons you have learned. (Pay close attention to the set of “Key Points,” helpful reminders at the end of most chapters.) These two simple tactics will facilitate, indeed expedite frequent reviews of key material later.