4 Technology Jobs Most Threatened by A.I.

The rise of artificial intelligence (A.I.) and machine learning may benefit the world in spectacular ways—but as many pundits have noted, such advances could come at significant cost.

If you’re someone like Elon Musk, you fear that the rise of machines will create a doomsday scenario right out of the “Terminator” films. But many economists and pundits think the actual consequences will involve casualties of a different sort: as increasingly intelligent systems take on more responsibilities traditionally left to humans, industries such as customer service and manufacturing could suffer massive job losses.

Earlier this year, a pair of Boston-area economists crunched some datasets and concluded that when robots are introduced, human wages decline. “Bottom line: Robots do replace workers,” read Bloomberg’s article on the study. “On the other hand, some industries that don’t automate end up losing workers anyway, because their costs are too high and their customers go elsewhere. For workers, robots are only part of the problem.”

With all that in mind, here are five tech jobs that could see the biggest impact from A.I.

Customer Service and Tech Support

In 2016, Facebook unveiled chatbots for Messenger. Over the past year, the social-networking giant has sought to improve the program, adding both features and a “bot store” modeled after traditional app storefronts. But that hasn’t necessarily improved the functionality of the bots themselves: According to a February article in The Information, bots on Facebook Messenger could fulfill only “about 30 percent of requests with human agents” during testing.

Despite that very public debate over the effectiveness of bots, other companies have plunged into the arena with their own toolkits and products. Microsoft, for example, has its own bot service, which includes a “bot builder” on GitHub.

Giant tech firms’ interest in bots means that the software will no doubt improve—and that could end up having a huge impact on customer service, currently the target of much bot functionality. In theory, a sophisticated bot could service the vast majority of service requests, negating the need for massive call centers stuffed with people. Tech support could end up affected, as well.


If you’ve been paying attention to artificial intelligence in the news, you’ve doubtlessly read stories about A.I. platforms beating out human champions at games such as Go. That is a very big deal, as it suggests software can execute effectively tasks that require creativity and improvisation.

And if you create software for a living, you know that creativity and improvisation are the bedrock of your profession. It’s easy to imagine a future in which A.I. and automation have taken over many of the tasks associated with building and maintaining software—at which point, at least a subset of developers may find their jobs at risk.


The maintenance of IT stacks is one of the most critical jobs in tech. As businesses tighten budgets and look for ways to save money, though, executives will begin asking for automated solutions. In the corporate view, the fewer bodies needed to maintain uptime, performance, and security, the better.

Startup firms seeking to merge A.I. research with enterprise security have attracted funding of late, including Darktrace, which recently raised $75 million. Google is also developing A.I. that can tackle malware and exploits in app stores—a vital service for any company looking to manage what its employees download.

If A.I. can handle security and other aspects of IT maintenance, it could reduce the overall need for sysadmins—forcing at least some tech pros currently in the role to specialize in order to stay ahead.

Advanced Manufacturing

In comments to Bloomberg last month, Apple CEO Tim Cook seemed upbeat about the place of advanced manufacturing in the U.S. economy. “The likelihood of robotics absorbing an ­assembly-type thing fairly quickly is high,” he said. “But in advanced manufacturing, there’s going to be a lot of jobs.”

Whereas the factories of yesteryear required hundreds of people working in shifts, modern ones are more like machines that build other machines. Tesla’s Gigafactory, an enormous facility in Nevada that will employ as many as 10,000 people at full capacity, also relies on an army of highly advanced robots.

That’s not to say that machines will entirely replace the human role in advanced manufacturing, but the increasing sophistication of automated platforms means that robots will only become more capable of churning out products faster than people. As with sysadmins, that means people who work in advanced manufacturing will only need to become more specialized if they want to succeed in the A.I.-influenced marketplace.

9 Responses to “4 Technology Jobs Most Threatened by A.I.”

  1. I could just laugh at such stories or rather say myths. Yes, every new technology or functional practice tries to automate parts of some jobs yet they create more jobs and further scope for advancements. Four areas mentioned here are have history and evolution of their own. Having worked in 3 of them, I can categorically say A.I. is nowhere close to replace or threaten these jobs. It would be like arguing if McDonald’s store automates its customer service jobs will it eliminate all customer service jobs in McDonald’s? No WAY. Few hourly paid jobs may be eliminated to start with but after losing business targets! everyone falls back. The very example in front of us is call centers by many companies that they brought back to onshore or even many advertised as “human touch” when you call customer service.
    One thing is for sure, A.I. and alike influences to change nature of functioning but they will also create new areas of jobs at the same time. During decades 70s and 80s workers were scared by many that computerization will eliminate jobs but computerization created many and other job areas. I would accept a defeat in my argument unless and until a robot mimics an intelligent and learned human being. That day is very far from reality at the least 50-100 years, to say the least. We must often look for practical, facts and data based analysis and sometimes should listen and learn from great thinkers like Prof. Michio Kaku and ilk.

    • Its easyvyo say robots will not take over the work
      In the last thirty years of my trade I have seen people leave engineering due to more and more CNC machines.
      OK its not that bad you say but I see in five to ten years they will flip burgers or do the washing. Cook meals as a pro drive cars. When that happens something needs to be done to safeguard against retaliation from the displaced workforce or a time of total unrest will follow

  2. Michael

    It seems everybody is so afraid of AI taking over a lot of jobs. Maybe it will happen, but I doubt it. On the other hand, one of the greatest job killers of all time was the broad introduction of plastic, and it seems that no one ever noticed how many jobs has been killed by plastic. Just imagine if everything you see around you made of plastic would have been made out of other materials like metal, wood, fabric or leather – all of them need much more manual work. How many plastic shoes can a single worker get out of an injection molding machine a day? And how many out of non-plasic materials!?
    Forget about AI as job killer, we have seen much worse than that.

    • I agree. Every time I see “AI is going to replace you” I see “Visual programming will replace you” that was so prevalent about a decade ago. AI, and all software, is only as good as those who wrote it.

  3. Garrett

    I don’t see AI taking over software development for a long time due to there being such dynamic requirements. Games like chess and go have a specific set of rules, which are easy for an AI to follow and improvise within. Every specialized application has shifting requirements, some of which may not be clear at the start of the project (and some of which may not be clear until the project is finished). AI still does terrible with vague requirements, and I doubt the ability to predict non-existent requirements is in a much better state. For an AI to even begin to replace a software developer, it would have to have a HIGHLY customizable set of metrics that someone (likely with coding experience) could change on a per project basis, and the AI would need time to train on those metrics to find the best solutions. Rinse and repeat every single time a project requirement changes, which is rather frequent due to shifting customer needs and feedback. Point is, even if an AI can automate the actual generation of code or automate the compilation process, AIs will practically be people by the time they have the insight and fluidity to autonomously replace software engineers.

  4. Since when does AI beating somebody at a game qualify as “suggests software can execute effectively tasks that require creativity and improvisation”. All games, and I mean ALL, have a set series of rules and a finite total of moves possible at any given point in time. Throw enough computing power at it and AI could predict the ending of a game from the very first move. AI is not being creative at all, it is doing what it has always done, executing a series of pre-programmed algorithms and calculating the best most probable move. Now, give the AI the rule book, tell it to read it and learn the game, than start beating humans and maybe I can see it as creativity and improvisation. Until then the computer is just doing what the programmer tells the computer is possible and is just allowing the computer to pick the order. Programming will be around until the AI system can learn completely on its own from startup with no routines.

  5. bambib

    “If you’ve been paying attention to artificial intelligence in the news, you’ve doubtlessly read stories about A.I. platforms beating out human champions at games such as Go. ”

    Games like “Go” have readily-definable patterns which can be anticipated in programming. The nature of programming itself is fundamentally different – since the applications are essentially unbounded.
    For complexity, compare placement of a Go stone with the instruction set of a programming language. In Go, there is only one instruction. In most programming languages there are 100 or more. In Go, there are only two possible arguments to the instruction (the X-Y coordinates of the stone’s placement). In any modern programming language, there are an unlimited number of possible arguments. A game of Go is deterministic and the number of instructions in a program (moves in a game) are strictly bounded to a number that’s probably less than a couple hundred. A computer program can be of any length, and routinely exceeds hundreds of instructions.

    We already have IDEs that provide code blocks to aid programmers, and which prompt for code endings, but that’s not the same as an autonomous creation of code. The Java language can be largely viewed as stringing together modules of pre-written code to achieve a result – but until desired outcomes are homogenized, the creativity of programmers will still be required to write software.

  6. wge300

    I am 62. In the course of my adult working life (approx 40 years) I have watched technology get better and jobs go away as a “better mouse trap” has replaced the working person. In Broadcast Radio, I have watched people get replaced by better and better automation systems. As the technology that transmitted the signal got better (more reliable and stable) and became more computer based, I watched the need for a full time engineer at radio stations go away. And it has not been just broadcasting. How many of you remember ever seeing a caboose at the end of a train? OK, people in Oregon don’t count. The brakeman who once rode in the caboose and monitor the air brake system is no more. Replaced by technology. From 1995 to 2013 I worked for AT&T / Lucent / Avaya in the technical support center. Me and my co-workers would take on average between 10 to 25 calls a shift per person. Now, following in the footsteps of Cisco, and putting everything that they could on line and into call in menus, Avaya was able to route a lot of the phone traffic away from live bodies to automated answers. When ever someone says A.I. they always think of robots doing the work. But very often it is as simple as a computer program (ever order anything from Amazon or Ebay?) It is all automated, and if you need to get with a live body to solve a problem, well you are about out of luck. Get yourself a sandwich and be prepared to be on hold for a while.