by Dan Tynan

The coming IT job apocalypse: Rise of the machines

May 28, 2019
Artificial IntelligenceBPM SystemsCareers

AI and process automation are taking on tech jobs once thought untouchable. Will you be replaced by a robot?

ai artificial intelligence job apocalypse
Credit: sara5 / Getty Images

In the war between machines and mankind, the machines have gained the upper hand.

It’s bad enough that computers can now beat us at chess, Jeopardy, and Go. Artificial intelligence-driven algorithms are now tackling jobs once considered the exclusive province of living, breathing bipeds. That includes doctors, lawyers, teachers, and, yes, IT professionals.

McKinsey estimates that roughly half of all work activities could be automated using today’s technology, and that up to 30 percent of global workers could be displaced by 2030. The jobs of millions more will be changed forever by AI.

But automation will also create new roles and opportunities that did not exist before. Whether those new jobs will be sufficient to replace the ones made obsolete is an open question.

Are you at risk of being replaced by an algorithm? And if so, what can you do about it?

Here’s what you need to know about our glorious robotic future.

Take this job and code it

As in virtually every industry, the IT jobs that will be automated first involve repetitive and often manual work that doesn’t require a lot of human discretion, notes Keith Strier, global and America’s advisory leader for AI at consulting giant EY.

“If you’re a sysadmin, or tier one tech support, or even in cybersecurity but your primary job is to look for certain signals and indicators, your jobs are up for grabs,” he says.

According to researchers at Oxford University and the Kellogg School of Management at Northwestern University, database administrators have a 39 percent chance of having their jobs automated. If you’re an IT operations tech, that number rises to 78 percent.

Those numbers will also vary depending on where you live, says Dr. Hyejin Youn, assistant professor of management and organizations at Kellogg. The smaller the city, the more likely your job will be taken over by a machine.

“These job titles won’t become extinct,” explains Youn. “There will still be a need for humans in the occupation, but the tasks will change, and the number of people doing it will be much smaller.”

Because IT is continually asked to do more every year, usually without a huge boost in budget, Strier says IT departments are more likely to reassign employees to more advanced tasks and use automation to fill the gaps.

“It’s less about letting people go, and more about reduced hiring,” he says. “Doubling capacity without doubling headcount seems to be an increasingly popular way of looking at the savings automation creates.”

But what’s changing is the types of tasks that can automated, notes Forrest Brazeal, senior cloud architect for Trek10, a cloud consultancy. In a widely shared essay titled “The Creeping IT Apocalypse,” Brazeal wrote about the quiet decimation of low- and mid-level IT jobs brought about by the growth of cloud services and AI.

While the loss of jobs through automation has been a byproduct of technology advancement since the industrial revolution, Brazeal says this time is different.

“This is a sea change,” he says in a phone interview. “Entire disciplines will be going away. There will be much less call for Windows sysadmins, DBAs, and network engineers. That’s what a lot of people are missing, and it’s what I mean when I talk about the ‘creeping apocalypse.'”

Alexa, write me an application

One of the higher-level jobs that AI will soon take on is writing code. In fact, the quest to automate programming is already well underway.

In 2017, Google’s AutoML research project demonstrated that it could generate machine-learning software that’s sometimes more accurate than similar programs written by humans. It’s now available as a cloud-based service that allows developers with limited machine learning experience to train ML models.

Last year, computer scientists at Rice University unveiled BAYOU, an AI application that uses “Neural Sketch Learning” to generate code. After studying 100 million lines of Java on GitHub, the DARPA-funded tool is able to recognize high-level patterns in programs and recreate similar ones on demand. Enter a few keywords to tell BAYOU the kind of program you want to create, and it will spit out Java code to fit the bill.

AWS’s App Sync and Amplify “low-code/no-code” development automation services are another example of this, says Brazeal.

“The idea is to take most of the work out of creating a traditional back end, so you can spin it up and have it happen automatically with just a few lines of config,” he says.

Once that’s in place, he adds, you’ve eliminated an entire class of software developers. And the writing for many of the rest is clearly on the wall.

“We’re starting to see the beginning of ‘conversational programming’ — the ability to build services by saying, ‘Alexa, take these components and put them together to give me an application,'” he says. “We’re not there yet, but it’s something to keep an eye on.”

People first?

When executives talk about automation, they invariably say that relieving IT employees of boring and repetitive tasks frees them to take on more strategic roles and responsibilities. But when you ask what those new roles will look like, and how employees will make that transition, you tend to get a blank stare.

The fact is, few organizations have even thought about it, says Strier.

“The majority of these projects are being championed by a mid-level executive who’s been told to reduce overhead or improve customer service or some other goal,” he says. “They’ve not been empowered to worry about the future of their workforce. To them, that’s an HR issue. They truly believe automation will free up their workforce to do more important things, but no one does the work to figure out what that is.”

Strier says he has one client, a large telecommunications firm, that has thought through the implications of how automation will change what its employees do. That firm was forced to address these issues because it was heavily unionized.

“That enabled them to say to the unions, ‘Look, we’ve identified job classifications that will be more deeply impacted over the next three years. We can give notice to those employees and offer them an opportunity to think about retraining,'” he says. “That’s better than getting a letter on Monday saying, ‘We’re shutting down your department.'”

Implementing this kind of broad organizational change is not a trivial task, says Stanton Jones, director of research for ISG, a technology and research advisory firm.

“The vision many enterprises have is that they’re going to take that 30 percent of people’s tasks and repurpose them for something more important,” says Jones. “But doing that for hundreds or thousands of people is really hard work. I’m not saying it can’t be done — a small number of companies are really rethinking how their organizations can be run — but it requires they put people ahead of their cost savings and productivity goals. That’s pretty rare.”

Augmented, not replaced

It’s an article of faith that as jobs continue to be made obsolete by automation, new roles will emerge to replace them. And so it goes with AI, which will both dramatically change today’s jobs and create new ones that do not yet exist, notes Erik Brown, a senior director in West Monroe Partners’ technology practice.

“In a few years it will be hard to find a job that’s not augmented by AI,” he says. “Think about financial services and fraud detection, or risk exposure in investments. AI will be used by utility companies to predict how weather will impact energy demand, and by insurance companies to process claims. And there will be a lot of jobs in education, teaching people how to use their business knowledge to train algorithms.”

Likewise, Brown adds, network engineering jobs could evolve into roles that use AI to manage data centers more efficiently, as Google did with the DeepMind algorithms it developed to defeat a human champion at the Chinese game of Go.

One of the biggest sources of new jobs will be embedding AI into hardware such as robots or autonomous vehicles, says Strier.

“Integration of that software and hardware is very complex, and it won’t happen on its own,” he says. “So while you might have used AI to write some of the software, ultimately humans will need to do the integration, modeling, and testing of these complex hardware/software configurations.”

Newly emerging roles such as ITops data scientists, AIOps architects, and automation path designers will be created as a result of AI-driven automation, says Will Cappelli, CTO for EMEA at Moogsoft.

“The human side of IT will have to shift from observation to analysis, which is why professionals in these future positions will need skills involving mathematics and an end-to-end understanding of how modern IT systems behave,” he adds.

At the executive level, companies riding the AI wave are looking to hire data-savvy executives who combine business skills and analytics expertise, says Scott Snyder, a partner with Heidrick & Struggles, an executive search and consulting firm. 

“We place a lot of leadership positions like chief AI officer and chief data officer,” says Snyder. “We’re always looking for people with data-intensive backgrounds who can graft those skills with institutional or functional knowledge, such as HR, legal, or supply chain.”

Brave new jobs

AI is also likely to generate a raft of other jobs that are just barely visible on the horizon, says Amber Bouchard, director of talent acquisition for Maven Wave, a digital transformation consulting firm.

One of those roles could be “citizen data scientist,” she says. Such an employee would analyze data and extract insights, but without the need for an advanced degree in statistics — like a business analyst on steroids.

Another new role would be “neutral AI assurance expert,” a master coder who can detect potentially biased algorithms in complex machine-learning models. And once the impacts of AI start to become felt, companies may also want a chief ethics officer to oversee the moral implications of machine learning and AI in the workplace.

“Organizations will need a person who can partner with HR, senior management, and C-level executives to oversee the implementations of these new technologies,” Bouchard says. “Someone will have to erect virtual walls and help organizations navigate the waters of technological advancement.”

She adds that there will remain tens of millions of jobs that can’t be easily automated.

“There are many jobs that are susceptible to automation,” Bouchard says. “There are just as many jobs, especially within firms like ours, that require the judgment, social skills and hard-to-automate human capabilities that AI cannot take away.”

Robot wranglers

In the meantime, IT pros worried about being replaced by robots should think seriously about diversifying their skill sets and consider becoming full-stack engineers, says ISG’s Jones.

“People who can manage everything from the web server through middleware, the operating system, and even down to the virtual machine layer are in huge demand,” says Jones. “Organizations cannot find those people fast enough. But if you’re stuck on a single set of technologies, that’s going to be problematic.”

Jobs that consist of “undifferentiated heavy lifting” are always the first to go, adds Brazeal. The more generic the tasks you perform each day, the more likely you’ll be replaced by code. Developing expertise in areas that add bottom-line value to the company are the best ways to ensure job security.

And while you won’t need to become an AI expert, deep familiarity with the available AI solutions will become increasingly necessary.

“The skill sets the tech department will need to remain relevant won’t be building AI systems,” says Strier. “They’ll need to be experts on the different solutions in the field, the strategic use of this technology, and how to integrate third-party AI services into their operations.”

In other words, he adds, tech pros won’t necessarily need to know how to build a facial recognition algorithm, but they will need to know how to pick the right one for their company.

That’s something robots can’t do… yet.