Artificial intelligence seems to have permeated every conversation, especially with the advent of OpenAI’s creations, ChatGPT and Dall-E, but for most of us, AI is still a bit fuzzy. This dual-volume edition—Power and Prediction and Prediction Machines—is among the finest books to understand how artificial intelligence and machine learning are changing the world around us. Prediction Machines, which was first published in 2018, has been updated to better explain the economics of artificial intelligence, while the new volumePower and Prediction dives into the disruptive economics of AI. The prediction machines the authors write of are both software and hardware, including robotics, which decouple prediction from other factors of the decision-making process. This decoupling, the authors—all professors at the University of Toronto’s Rotman School of Management—argue, is what makes AI the most compelling technology to impact human civilization.
Also read: Best smartphones to make the leap from 4G to 5G
This is a game-changer as any decision-making process consists of several steps, including identifying all the solutions, identifying the dependencies and consequences of such solutions, putting weights on which solution might work in which circumstances, predicting the intended and unintended consequences of each solution and the decision that goes with it, and, last but not the least, having complete information about the entire universe of options to enable decision-making. Prediction is one part of the process and AI will separate that from the other components, making the decision far more comprehensive, quicker, and less erroneous.
It is a given that AI will automate most tasks, take on more cognitive tasks and do them well. The key question is how we humans will adapt. The more complicated areas relate to the use of judgment inside complex ecosystems such as government where what is said, what is meant, what is seen and what is unseen are completely different and often indiscernible even to the trained human eye. How artificial intelligence will select from among options which are fuzzy—and can mean both one thing and it’s opposite at the same time—will be a huge challenge for modern prediction machines.
The section on the “between times”, or the time before a massive breakthrough is used to its full potential and has maximum impact, is particularly enlightening. They explain “between times” with electricity, which was a breakthrough technology—Edison’s moment with the lightbulb happened in 1879, but even 20 years later only 3% of American households had electricity. After another 20 years, 50% of American households got electricity, and another 10 years passed before 90% became accustomed to it. This gap between the breakthrough and its use becoming ubiquitous and invisible is where we are right now with AI. Right now, we are using AI for “point solutions”, or to address particular pain-points in a unidimensional way (think, reading x-rays or autonomous vehicles), but the real transformation will come when we reach the “systems solution state”, when AI not only solves an issue but also changes the dependent system and procedures. This state, when AI’s use becomes second nature to us, is 10 to 20 years away.
The authors oversimplify the prediction problem by looking at only a single necessary prediction: What would a human do? While framing the problem this way may help an engineer move beyond a rules-based programming tree, far greater nuance and understanding is needed to provide workable rules. For instance, while talking about autonomous vehicles, this single question does not capture the complexity of a machine saving or taking lives. Replace the autonomous car with autonomous weapons and the problems become more inscrutable.
Both books, read together, deepen our understanding of the applications of artificial intelligence and machine learning in day-to-day life. The authors peep into an uncertain future and draw out the contours of possibility. Ethical challenges will remain, and there is no ready answer for the question of whether automation will create more jobs than they displace. Drawing on history, we could say that every revolution, be it the agrarian, the industrial or the information, led to the creation of newer, better jobs than those that they displaced. So, there is no reason to be pessimistic about the artificial intelligence revolution.
The author is an IAS officer and tweets at @srivatsakrishna
Also read: Siddhartha Mukherjee discusses the future of humans