Skip to content

25th Anniversary series: On the Brink of Machine Learning Revolution

September 2, 2015

The term “Big Data” has escaped the boundaries of the IT world. Even if people don’t know exactly what the term means, they’ve heard about the endless possibilities presented by Big Data and smart algorithms. Still, few people grasp how disruptive these developments are going to be.

Just a few decades ago, most people had never even seen a computer. Since then, we’ve gotten used to thinking of them as tools that help us do our jobs. We tend to see Big Data in that same light: a tool to help us know more, understand more, and make better decisions. But the truly disruptive part of Big Data lies in what’s called “machine learning” (ML) – the ability of computers to not just analyze data, but to use it to make decisions and predictions.

Better than human

Why is that disruptive? While we’ve been all too happy to turn over menial calculations to computers, we’ve regarded decision-making as a uniquely human skill. Not anymore. Automation has evolved from the menial to the meaningful, and these smart computers have the potential to replace us in jobs we always thought were “safe.” It’s almost a certainty that any narrowly-defined cognitive task can – and eventually will – be automated. What’s more, even experts in their field will be outperformed by automated solutions. Nearly any industry can be revolutionized by scalable systems, because unlike their human counterparts, they’re able to provide help 24/7, with unlimited service capability.

That concept isn’t as far-fetched as it might seem. We’ve all heard about things like computers that beat humans in chess,[i] operate self-driving cars,[ii] and even simulate convincing human chats.[iii] But those are just the things that make headlines. Most of machine learning applications are already being used in more mundane (and therefore less newsworthy) business processes.

The future is already here – it’s just not evenly distributed

Marketing is just one example. From lead generation and advertising to upselling, cross-selling, coupons, discounts, and customer retention, marketing tasks are rapidly being taken over[iv] by computers. In retail, ML-based demand prediction is being used for things like pricing, assortment optimization, and maintaining optimal stock levels. In equipment-heavy industries, machine learning is the foundation for predictive maintenance.[v] Algorithms analyze historical data about equipment usage, defects, and upkeep, in union with 3rd party data, such as past weather conditions and future forecasts, to predict equipment failures before they happen, so the equipment can be serviced or replaced with no operational disruptions. Speech recognition and sentiment analysis are taking over customer support. Image recognition has numerous applications in industries from healthcare[vi] to smart cities.[vii]

And these machines aren’t going to stop learning any time soon. Not only can they process more information than would be possible for a human, they’re also less prone to human mistakes and biases. All of that combines to deliver personalized service on a scale never before seen. And they’re just going to continue to get better, faster, and cheaper.

This won’t happen overnight, of course. Early adopters will be the ones who expect to see the biggest benefit, which is why, in the grand scheme of things, we’ll see an omnipresence of self-driving heavy trucks[viii] before self-driving cars get widely-adopted, for instance. In addition, many industries still need to accumulate the data these machines need in order to learn. That’s why marketing and e-commerce are leading the pack: there’s plenty of data, and you can easily conduct an experimental campaign to measure the benefits. Most brick-and-mortar businesses, on the other hand, still have some work to do.

As part of this revolution, we expect that new and currently unavailable services will flourish. Consider, for instance, the benefits that health monitoring and prediction systems could bring to personalized healthcare by analyzing genomes. Patients in all demographics would be given access to the kind of high standard healthcare that even the privileged-few today cannot obtain.

It’s closer than you might expect

So, to sum it up, here’s our prediction.

By 2040 – and maybe sooner – algorithms will be doing the work in all business processes where:

  • We know exactly what we want to improve and can measure it
  • We have enough data
  • We can experiment
  • We can take automated action

Think things can’t possibly change that much in 25 years? The folks at Google would tell you you’re wrong. Fifteen years ago, nobody had even heard of Google. Now the company’s name is a verb, and they’re disrupting industries far beyond the internet, like banking,[ix] travel,[x] and even car manufacturing.[xi]

From science fiction to “must have

Machine learning is already here, and it’s moving from “innovative” to “must have” at a breath-taking pace. McKinsey, in a recently published article entitled “An Executive’s Guide to Machine Learning,[xii] already describes it “as a mainstream management tool.”

As the amount of data we generate grows on an exponential scale (90% of the world’s data was created in just the past two years),[xiii] so does computing power (see Moore’s law).[xiv] This means that machines will improve their capabilities faster than ever before, while at the same time becoming cheaper. That’s what “disruption” is all about – dramatic changes that happen extremely fast.

Remember 1950s science fiction? You might recall how our futuristic world was run by robot-doctors, robot-consultants, and electronic brain-navigators. As it turns out, these early representations were not so far-fetched. In fact, however eerie it may be, these depictions are rather accurate precursors for how technology influences life and business today.

The New Industrial Revolution

It’s difficult to overestimate the potential disruptive impact of the coming changes. Most likely, the “ML Revolution” will be comparable to the Industrial Revolution, which left a significant percentage of British workers unemployed – up to 75% in some trades! – but also led to unprecedented growth in population, productivity, and standards of living. And, just as in the Industrial Revolution, we’ll see many skilled workers – educated, middle-class citizens of developed countries – losing their jobs to machines.

McKinsey’s guide, mentioned above, puts it quite bluntly: “Change is coming (and data are generated) so quickly that human-in-the-loop involvement in all decision making is rapidly becoming impractical.” Both whole industries and their surrounding ecosystems will be reformed. Think about the implications of self-driving trucks. What’s going to happen in small U.S. towns when there are no truck drivers to stay in hotels, eat in cafes, and shop in the roadside stores? Moreover, who would purchase a car if, in mere minutes, an incredibly cheap, automated vehicle could be rented instead?

Society, redefined

Optimists note that, after the initial shock of the Industrial Revolution, unemployment rates eventually fell back to the normal levels. Many people found new jobs – ones that didn’t even exist before – so we can hope for that same outcome, that we’ll just invent something new to do.[xv] But even if that eventually happens, society will still have to adapt to skyrocketing unemployment during the initial phase. Liberals already assume that “the only real option is redistribution and a lot of it,[xvi] and that the only meaningful question is the form of such redistribution.[xvii] More conservative ones insist[xviii] that, as Margaret Thatcher said, “The problem with socialism is that you eventually run out of other people’s money.” But that response doesn’t provide solutions for a society where half of the population is unemployed, and we should probably expect that the pressure on central governments to move toward the redistribution of wealth will only grow. On the bright side, chances are that quality of life will significantly improve, even for low-income households. The coming automation will make many things, even beyond basic needs like food and shelter, much more affordable. Personalized service, from healthcare to shopping, will become a commodity.

There’s yet another impact to consider: cultural change. For most people, their job is not just a way to put food on the table. It also gives them a sense of purpose and belonging, of being needed and appreciated. Our culture innately believes that idle hands are the devil’s workshop, and you’re not to be respected if you’re not working hard. So, when we no longer need to work so much, will we be able to change our attitudes accordingly? How will society judge those who are not involved in productive work? What new activities will be supported and respected? What will your profession be in 2040? Data Science? Pottery? Or some entirely new job you can’t even conceive of today?

There are a lot of questions yet to be answered, and there’s not much time left to find the answers. Maybe we don’t know exactly what this new world will look like and how it will function, but it’s time to start thinking about it.

Written by Jane Zavalishina, CEO, Yandex Data Factory and Alexander Khaytin, COO, Yandex Data Factory;

There are many excellent guides to major trends that will affect us all over the next decades. For guidance, see . What we have tried to do in this blog sequence is to highlight a specific emerging change from the many, and to explore some of the potential impacts.  We welcome thoughts on other drivers of change or more impacts of the ones we have highlighted

If you enjoyed this blog, sign up for our monthly newsletter at and/or browse our website at



















No comments yet

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 )

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

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: