Today, at the ML School for Business Schools, organized by BigML, I gave a talk about the importance of teaching machine learning in business schools from a horizontal rather than vertical perspective, introducing the tool in each area of study, rather than placing it within a specific course, and how educational models can influence the adoption of machine learning at the corporate level.
Machine learning is, in many ways, an extremely disruptive technology. The idea of building algorithms capable of automatically improving through experience is not new as a discipline, in fact, it dates back to the 1960s, but it experienced a long winter and strong disinterest due, fundamentally, to the lack of resources to enable the levels of data processing that the discipline required. From an academic point of view, it reminds me of what happened with two of the most representative subjects of my experience as a teacher: spreadsheets and the internet: a period of almost exclusive use by specialized groups and unfriendly usage models prevented the spread of innovation, followed by a period in which the development of progressively simpler interfaces and tools brought their use to an increasingly wider public.
If one thing is clear to me, it’s the horizontal nature of machine learning: in companies, we find use cases in practically all areas: operations, logistics, marketing, finance, information technology and systems, etc. I was also clear about the horizontal nature of spreadsheets and the internet: their use should be exclusive, respectively, to areas such as finance or technology, but it was obvious they would be extended to all functional areas. At my institution, IE Business School, we use updated versions of some of the cases I wrote at the time about spreadsheets, which had to do with processes of all kinds, from relatively complex financial calculations to the automation of processes in marketing or customer service. The spreadsheet has long been a universal tool and a lingua franca in business.
I had a similar experience with the internet: it was introduced in business schools in the area of information systems and technologies, and for several years, everything to do with the web fell into this area. It didn’t matter whether the case was strategy, marketing or operations: if the company had a .com in its name or its operations took place online, it was up to those of us working in that area — we’re talking about the late ’90s or early ’00s, when Andy Grove’s phrase that “all companies will be internet companies” still raised smiles among skeptics.
That top-down approach, in my view, has led to a long delay in the adoption of these technologies at the corporate level. For many years, companies were unable to extract full value from tools such as spreadsheets or the web, simply because they mistakenly assigned their use to specific functional areas, be it finance or technology. Many years later, when the adoption curve included more people and institutions, we understood that in reality, these were tools that could — and should — be used for absolutely everything, and that there were many advantages derived from doing so.
It’s now the same story with machine learning, which is being introduced in business schools either as something related to technology, or as specific courses that usually appear first as electives or workshops. This approach not only limits its use and the transfer of knowledge to corporate environments, but also complicates its learning with a fundamental problem: we tend to think that in order to use machine learning, our students must learn to build the necessary tools for it.
This is like asking someone who wants to use a spreadsheet to learn how to program one, as was the case in those very heavy first sessions during the early days of the internet when we would spend the first half hour on how to connect (I still remember the creaking sound of the modem’s handshake), or how we now spend several expensive sessions teaching our students the basics of Python or R so that they can do things that, when they need them, they will do simply with a few clicks on menus and buttons. As Cassie Kozyrkov rightly says in one of the best articles I’ve read about the field, it’s like hiring expert electrical engineers to build ovens in order to set up a bakery.
Quite simply, there is no point teaching students to make their own tools, much less restricting machine learning to the realm of technology, or worse, to their own, to a kind of watertight compartment. Expecting students to go to the companies that hire them and dedicate themselves to programming algorithms in Python or R, is a huge source of frustration when, months later, it’s shown that the algorithms they have created are practically impossible to put into production. The latter simply prevents students who should know how to use and apply machine learning to practically anything that generates data and moves from being interested in it, because they interpret it as something that is specific to a particular functional area.
We now find ourselves in a latter-day version of the old saying: the cobbler’s children always go barefoot. The same business schools that design and deliver in-company courses of all kinds are apparently incapable of designing in-company courses for themselves. Otherwise, they could easily solve the problem: teach machine learning to professors who teach operations, finance, marketing, entrepreneurship, or any other functional area. Supervise and design machine learning applications with them for their respective areas, so that students can experience machine learning for what it is: a very disruptive horizontal technology, capable of being applied to problems of all kinds, in all fields, by practically anyone with a minimum knowledge of statistics.
In short, you don’t have to be a data scientists to learn how to use machine learning; far from it. You do need to understand a little about the specific economics that data generates, but not much more. Believing so is, again, a source of frustration and intimidates people. It is causing our students to come away thinking that machine learning is the stuff of Terminator, instead of understanding that it is something that will affect the daily lives of their customers and themselves. In other words, a mistake that is holding back the arrival of a tool with enormous potential from companies of all types, of all sizes, in all industries. A mistake that we have already made with other tools, many times before. But above all, a mistake that can easily avoided.
This post was previously published on Enrique Dans.
You Might Also Like These From The Good Men Project
|Compliments Men Want to Hear More Often||Relationships Aren’t Easy, But They’re Worth It||The One Thing Men Want More Than Sex||..A Man’s Kiss Tells You Everything|
Join The Good Men Project as a Premium Member today.
All Premium Members get to view The Good Men Project with NO ADS.
A $50 annual membership gives you an all access pass. You can be a part of every call, group, class and community.
A $25 annual membership gives you access to one class, one Social Interest group and our online communities.
A $12 annual membership gives you access to our Friday calls with the publisher, our online community.
Register New Account
Need more info? A complete list of benefits is here.
Photo credit: iStock