Understanding the Power of Prediction

I recently read Prophecy: Prediction, Power, and the Fight for the Future, from Ancient Oracles to AI. The book’s author is Carissa Véliz, associate professor at the Faculty of Philosophy and the Institute for Ethics in AI, as well as a tutorial fellow at Hertford College, at the University of Oxford. Before Prophecy, Véliz wrote the well-received Privacy is Power.

I haven’t read Privacy is Power yet, but I strongly recommend Prophecy. Véliz entertains while edifying, taking us through the history of prediction, from the divination and prophecy of antiquity to our modern, allegedly data-driven approaches. You doubtless noticed that I included the word allegedly in that last sentence. That’s because Véliz makes a compelling case that data doesn’t drive anything, certainly not contemporary predictions. People frequently use data as a historical foundation for predictions, but the data does not have volition or a driver’s license. We use it as we please, and as it pleases us.

The salient point that Véliz wants us to take from Prophecy is that predictions are often veiled power flexes. The future is unwritten, as we know, but a plausible prediction from a high-profile industry luminary can nudge the pen hand of fate by persuading us of its inevitability. This is what can be described as an attempt to engineer a self-fulfilling prophecy.

Not all predictions are power plays. I was, as you’ll recall, in the prediction business for about 15 years of my 33-year career in what was, and perhaps still is, information technology. My first few years in technology were in market research and forecasting, and then I returned to the profession, at IDC, for more than a decade on the back nine of my career.

When I prepared and published forecasts, I can assure you, I was not exercising power. Together with my colleagues, I was simply trying to get things right, as in accurate. Early on, I realized that predictions were a fool’s errand. You were destined to miss the target; the only open question was the margin of inaccuracy.

So, we did our best to mitigate the risk of getting things horribly wrong. We built robust models on the firm foundations of credible assumptions and rigorous methodology. We considered all the things that could go right, but especially all the things that could go wrong. As market analysts, we tried to account for all the factors and variables, but there were some (COVID-19, anyone?) that you could never adequately anticipate. At IDC, we were fortunate enough to have economists on staff; they helped us account for macroeconomic and geopolitical shifts that were well outside the scope of our technical domains. We did our best with what we had, but we knew fate would ultimately puncture our pretensions.

The Power of Prediction

In Prophecy, Véliz offers some wise counsel: if you have to predict, predict smartly. A supporting point follows: “The closer you remain to the present, the more likely you are to be right.” She speaks the truth. I know from cruel experience that the further out a forecast trundles into the wilderness, the more exposed it becomes to threats that can maim or kill it. You can only see so far, and the same contingency that keeps our lives interesting also thwarts our expectations, especially if those expectations are, in retrospect, unreasonably optimistic. We need optimism as fuel for aspiration and vitality, but if not held in reserve, it can put on us the hubris express, also called the road to ruin.

Véliz also enjoins us to predict ethically if we must predict. If we don’t need to predict, we shouldn’t bother. It is better, for reasons she explains cogently in the book, to prepare thoroughly for uncertainty, which is, alas, the only future certainty. She offers other prescriptions toward the end of the book, all of which are eminently sensible.

My job, early and late in my career, was to produce forecasts and predictions. I understood implicitly that I carried great responsibility, though I did not perceive any exercise of power. Other people, who aren’t strictly in the predictions business, make self-serving prophecies, and Véliz advises us to be wary of them. She writes the following:

“We live such fast-paced lives that half of the time we don’t even identify predictions as such. We listen to the assertions of prophets as if they were describing the world, instead of prescribing it.
The first task is to realize that someone is offering you a prophecy, and not a truth statement. The second is to ask questions.
What is this person doing with this prediction? Do you have good reasons to trust them? Are they pursuing knowledge or profit? What is the prediction pushing you to do and who will benefit from it? Is the prediction based on data? What kind of data? Who collected the data and why? What are the values that drove the data collection?”

Profiting Over Prophecy

These are all great questions and laudable cautions. When my colleagues and I put together our forecasts, we could cite our historical data and its provenance, openly declare our motives (it was our job), and we were transparent about the purposes and objectives that the data was meant to serve.

Conversely, when Elon Musk adds to his innumerable predictions, or Mark Zuckerberg makes his latest prediction about the coming ubiquity of AI agents, or Sam Altman declares that we are entering the singularity, how should we react?

Well, Véliz has given us some qualifying questions to test the mettle of the tech industry’s soothsayers. In these cases, you will find that the predictions are the power plays that Véliz warns us against. These predictions are speculative, not factual, prescriptive rather than descriptive. The prophets want you to accept their prescription and to collude with them in prodding it toward self-fulfilling fruition. Their prophecies are not supported by data, nor are they scaffolded by sturdy methodology or explicitly stated assumptions. You should treat these predictions with well-earned skepticism. The motive behind them is profit, not prophecy.

Predictions aren’t always disguised projections of power, but often they are. By asking the right questions, by approaching predictions critically, you can live better in the present, prepare for the future, and protect your own interests.