From Prediction to Transformation

Article Source: Harvard Business Review Magazine, November-December 2022
Publication Date:
Time to Read: 3 minute read
Written By:

 Ajay  Agrawal

Ajay Agrawal

 Avi Goldfarb

Avi Goldfarb



Artificial intelligence (AI) systems improve the quality of business decisions. Early applications of AI are not transformative. However, AI can transform business processes entirely by changing patterns of decision-making.


Policy Relevance:

AI has ripple effects that will ultimately transform all business processes.


Key Takeaways:
  • The key contribution of AI to business is to improve the quality of decisions, and businesses today make more decisions than ever before; in 1960, about six percent of jobs required workers to engage in tasks such as problem-solving and strategizing, but by 2018 about 34 percent of jobs required those skills.
  • Emirates Team New Zealand integrated AI into the design of the team’s sailing maneuvers, helping the team win the America's Cup sailing race.
    • In the time human sailors would take to run a few simulations, the AI could run thousands.
    • The AI suggested approaches that struck human sailors as counterintuitive, but that worked well on the water.
  • Early applications of AI reduce the cost and improve the accuracy of predictions that humans were already making, but these applications do not transform existing businesses.
  • AI-based systems can transform industries by changing how decisions are made.
  • Like other decision-makers, experienced taxi drivers 1) make predictions using data, and then 2) exercise judgment, taking account of factors not easily reduced to data; AI now analyzes the data to provide a route, leaving ride-share drivers free to focus on subjective factors that improve the comfort of a ride.
  • AI can concentrate decision-making or change who controls key decisions.
    • AI resume scanning lets one human resources person do the work formerly done by large teams.
    • AI X-ray analysis shifts power from the radiologist, whose judgment is no longer needed for diagnosis, to the patient's physician.
  • Use of AI by one business affects others in the supply chain; when a restaurant that regularly orders 100 units of a food per week changes to variable orders depending on an AI's predictions of diner demand, his supplier will also need to use AI to predict variable restaurant demand.
  • An ambulance system requires a central dispatcher to assign an ambulance to each call, to avoid a situation where none or too many arrive on the scene; this is an assignment problem, a type of problem that requires a communications system so that individual decisions about resource use are coordinated.
  • AI will help solve assignment problems by helping firms balancing coordination and modularity.
    • Coordination helps firms manage the lack of reliability associated with AI’s constantly changing recommendations.
    • Modularity insulates some decisions from AI’s ripple effects, reducing the need for reliability.
  • AI will ultimately have transformative effects across the economy, like electricity and personal computing; because of ripple effects, AI will ultimately transform all business processes.



Joshua Gans

About Joshua Gans

Joshua Gans is the Jeffrey S. Skoll Chair of Technical Innovation and Entrepreneurship, as well as Professor of Strategic Management, at the University of Toronto Rotman School of Management. Previously he was a professor at the Melbourne Business School, University of Melbourne, and at the School of Economics, University of New South Wales.