The landscape of the modern workplace is likely to be totally transformed in the coming years by AI.
It’s hard to talk about the future of work without artificial intelligence (AI) being mentioned in the same breath. Hailed by leading AI researcher and Stanford adjunct professor Andrew Ng as “the new electricity”, the proliferation of AI will send seismic waves of change through our society.
The workplace will not be exempt from this. So, what kind of practical changes can the workers of the future expect to arrive in the coming years?
Science Foundation Ireland’s Adapt Centre is focused on shepherding in the new technological age. Its far-reaching research covers topics such as cognitive systems, deep learning and machine learning, virtual and augmented reality, and, of course, AI.
AIs will help work, not steal your job
Both Lawless and Conlan agree that deep-seated anxieties about AIs stealing jobs are unfounded. If anything, AIs are likely to provide a boon to both your professional life and the economy at large.
“Individual workers may have a suite of AI agents that act on their behalf, typically to perform mundane tasks,” Conlan said. “For example, one AI agent may automatically respond to simple email requests and another may manage your appointments.”
According to Lawless, AIs are creating “a whole slew of professions that don’t exist”. The revered STEM role of data scientist barely existed a few decades ago. Its current ubiquity belies that fact.
In turn, some jobs probably will be wiped out. In 2013, researchers from Oxford laid out the types of roles most susceptible to computerisation. In truth, people in these professions do have some reason to be alarmed.
“This is the nature of progress; I’m sure that the equine industry was outraged at the development of the automobile,” said Lawless. “However, I believe, and Gartner agrees with me, that AI will help to create more jobs than it replaces.”
Chatbots are making lives easier
The rise of chatbots is a particular trend that Conlan notes. He claims that their implementation will reduce human-to-human interaction for people who work in professions such as customer service or even sales. This is “what many businesses are moving towards”.
If you have clicked into the ‘chat’ function on a company website, you have likely been speaking to an AI. Not that this is news to anyone – in their current, rudimentary form, they are easy to spot.
“[Today’s chatbots] attempt to match certain keywords from what you type and then follow a pre-scanned script to guide you to what they think the answer is,” Conlan explained. “They are very task-oriented and still tend to hand-written [queries] as glorified if-then-else programmatic statements.”
Still, even if imperfect, chatbots are good at fielding low-level queries, things a human employee may go into autopilot while doing anyway. Conlan is optimistic that saving this mental energy will greatly benefit both workers and enterprises. “This will free up employees to do the things we’re best at: being creative, being inspired, handling grey areas, communicating with others.”
You’ll help AIs work, too
The relationship between AIs and workers is set to be a symbiotic one. Though incredibly powerful and brimming with potential, these machines and mechanisms are bootless without our input, so Conlan says.
Many of the most popular techniques, such as machine learning, are data-driven. They’re powerful techniques, sure, but they “typically require large amounts of indicative data to learn”.
Conlan continued: “Much of this data is the result of human decisions, which are influenced by sometimes difficult-to-quantify things such as compassion or bias. The AI learning will absorb all of this, but tends to be very task-focused and can be brittle when required to act outside of what it has learned.”
An AI programmed to approve mortgages can be trained based on the reams of data created over the past five, even 10 years. It will perform well until the market inevitably undergoes significant change and its knowledge base becomes less relevant. “Humans are very much needed to generate the data upon which the data-driven AI will learn.”
The AIs of the future will need to be personalised and controllable to fulfil the complex needs of each enterprise. In order to do this, humans will have to interact almost constantly. “Currently, only a small proportion of our lives are digitised in a form that data-driven AI could learn from. As such, it can only act expertly in constrained ways,” said Conlan.
To facilitate this, employees may have to change the way they work or submit their work to greater levels of monitoring so AIs can have a constant stream of data.
The AI workplace has eyes
Monitoring in the workplace is a definite bone of contention, and one that Lawless maintains isn’t given enough air time in the AI discourse on future of work.
In particular, so Lawless said, little attention is given to how AI will influence the office as an environment. He believes that the AI-enabled office of the future is one where the walls – and every other thing employees interact with – will have eyes.
“AI will allow the constant monitoring – perhaps bordering on surveillance – of the workplace and the employees in the workplace. This has many potential downsides with regard to employee freedoms and privacy, but has lots of benefits both for the employer and employee … It will lead to a more effective and productive work environment.”
AIs could be used to monitor how employees interact with and move around their working space “to identify the best office layout to support collaboration and teamwork”. Though the privacy implications are disquieting, it could be a great way to discover trends that can steer office design.
Smart changes could, if implemented correctly, increase productivity and decrease employee stress. These are good things, but the ramifications shouldn’t be sniffed at.
It isn’t a question of if AI becomes a core part of your working existence, it’s when. This is why it’s good to understand the more granular elements of how your job, career and day-to-day workflow could be impacted.