Aug 2, 2019

How Machine Learning is Transforming the Way We Work

By Jennifer Rivera

When the steam engine was invented and put to work in the 18th century, it led to a transformation in the global economy in ways that had never been seen before: a veritable revolution. Other major innovations (electricity, the internet, etc.) also created massive changes in the way we work through the years. But not since the steam engine have we seen a technology with the power to create a paradigm shift in the way we work.

With the emergence and broad adoption of machine learning, a new paradigm shift is upon us. ML enabling computers to use algorithms to perform a certain task without the need of explicit instructions has brought about significant changes that have led to groundbreaking innovations in facial recognition, natural language processing, and computer vision -- not unlike how the steam engine brought about new modes of mass transportation and manufacturing. While certain tasks will be automated due to ML, the technology will also enable us to spend more time on thoughtful and creative work.

The capabilities of machine learning are already being used for fraud detection, media recommendation engines, and even financial analysis. Other applications, like next-generation medical diagnoses, are on the way and already being pioneered. This is particularly important, given ML’s ability to recognize bias.

With so many uses on the horizon, it bears taking a minute to wonder how the innovations enabled by ML might affect professions across industries. Let’s consider how ML could affect certain jobs, tasks, and work environments as soon as 2020.

When Online Tasks Become Automatic

Across professions and industries, work these days happens online. While it might be impossible to calculate just how much work happens online, it’s probably safe to say that most jobs and careers are dependent upon staying connected to some extent -- even when you consider the smaller details of scheduling appointments, sending emails, and collecting payments.

In stark contrast to jobs that require a physical presence or manual labor, jobs and job-related activities that take place mostly online are primed to be affected by machine learning. Administrative positions -- especially roles involving data entry -- are most certainly in the path of disruption of machine learning.

Businesses across industries often deal with the chaos and confusion that are wrought from having duplicate, missing, or unclean data. While sometimes this is due to the fact that sometimes pieces of data are missing, far too often human error is to blame. ML enables businesses to correct data inconsistencies. In the case of data that’s missing, ML offers data systems remarkably accurate “guesses” that can fill in the blank when information is absent.

Nonetheless, those that work in roles with tasks that might be better handled ML have no need to fear. In the case of administrative roles, there will still be a need for a ‘steward’ to monitor the activity being performed by ML. Further tasks such as returning phone calls and creative problem solving will be best performed by humans.

Inverse Reinforcement Learning Jobs That Could Soon Change

While certain jobs that require a physical presence and/or manual labor are safe from ML’s disruption for the foreseeable future, certain tasks in th is realm are not. Jobs in the transportation, security, and food preparation, in particular, are most certainly in the path of disruption. We are probably a bit further from autonomous vehicles being a reality than hyperbolic sentiment would have had us believe just a few years ago, but it is on the horizon. And once it does, companies like Uber and Lyft as well as public transportation systems are set for a sea change.

Meanwhile, food preparation at places like fast food restaurants could soon be handed over the machine learning. This is not to say these restaurants won’t still need human employees -- people are still much better at interfacing with customers -- but rather, the roles will evolve considerably.

Also take into account various security roles -- especially the ones that require checking and confirming identification. With the advances that have been made in facial recognition, it’s likely that ML has already surpassed even the best trained and experienced security professional at spotting cases of false identification.

But as with other physical presence-dependent jobs, the human element will still be valuable, in this case on the technical side of things. Ensuring that systems are running correctly as well as handling situations that require face to face interaction are examples of tasks that we’ll still need human security personnel for.

Identifying versus Analyzing Relevant Data

But while thoughtful analysis is still dependent upon a thoughtful person to be done most effectively, isolating the data needed to perform the analysis is something that ML can handle in fractions of a time that it takes even the most efficient, intelligent, and experienced human to perform.

This is good news for many professionals: For many, analytical skills, deductive reasoning, and even conveying this information to the relevant stakeholders are skills that come naturally to many people. (At the very least, they are skills that can be mastered through learning and practice.) Collecting, organizing, and culling the data that needs to be interpreted, meanwhile, is the challenging part that is time consuming, thus delaying the crucial analysis.

Consider a profession like a dermatologist, we can deduce that ML might be quicker at recognizing that a patient has a rare form of skin cancer that requires aggressive treatment. The dermatologist, meanwhile, will still be needed to deliver the news in a way that’s both intelligible and compassionate. Certain pieces of information simply need to be delivered by a fellow human.

Quick Decisions versus Chains of Reasoning

In most knowledge-based professions, there exists a constant (if invisible) dichotomy between tasks that require quick decisions versus chains of reasoning. As it turns out, ML is more effective at the former as opposed to the latter.

To return to a previously-mentioned task, consider appointment scheduling: For some people, something as simple as putting a meeting on the calendar with a client can be more deliberative and time-consuming than it’s meant to be. Running through one’s head of unscheduled and anticipated occurrences can slow us down and create undue burdens. This is the kind of thing, on a task-level, that ML could take off of our hands in short order.

Meanwhile, tasks that require thoughtful analysis and reasoning -- such as deciding on which vendor to choose and the best price level to negotiate, or perhaps figuring out how to diffuse a tense situation with a customer -- will continue to require an intelligent and strategic human in order to be best done effectively.

ML will be behind the scenes, becoming more efficient so individuals can spend more time with people. At the end of the day, the right interfaces for humans are humans, and not screens or keywords.

Upon a snap judgment, it’s easy to be wary thinking about how ML is going to change our working lives. But when we take a step further back to consider work on a task-by-task basis -- considering both the rote and mundane aspects of our work as well as the more thoughtful, creative, and empathetic areas -- a clearer and considerably brighter picture of work in 2020 begins to take hold: one in which we will spend our working days doing the things that make us human, and not machines. 

jennifer-rivera

By Jennifer Rivera

Jennifer Rivera is part of the editorial team at Publicize and Innovation Outreach. Earlier she led newsrooms at multiple international publications. 

<Return