Yes, AI will upend many jobs, just not the ones you imagined By Aron Heller | Nokia, Monday, 02 Dec 2024

One of the more widespread concerns of the looming era of artificial intelligence is how it will impact the traditional job market. Fears have been rampant on how it could make many blue-collar jobs obsolete, with robots replacing those in professions that require manual labor and thus further increasing gaps between the haves and have-nots.

But a new comprehensive Nokia Bell Labs study challenges that prevailing narrative, finding that it is white-collar, highly technical jobs that will likely be those that face more disruption.

After analyzing more than 24,000 AI-related patents filed with the United States Patent and Trademark Office between 2015 and 2022, a team of England-based researchers has developed an Artificial Intelligence Impact (AII) score that assesses how closely an occupation’s tasks aligns with recent AI innovations.

Their findings are no less than staggering, defying the conventional wisdom that AI will mostly harm routine, low-skill jobs.

In fact, they found the among the least impacted professions were pile driver operators, dredge operators, butchers and graders of agricultural products. Alternatively, skilled professions such as cardiovascular technicians, sound engineers, animators and orthodontists were those most vulnerable to AI impact.

“We found that the key factor was not the level of skill or physicality involved in a job, but rather whether its tasks followed a structured, machine-readable format,” explained Daniele Quercia, Department Head of Social Dynamics at Nokia Bell Labs Cambridge (UK). “Jobs requiring precise sequences and predictable outputs—such as those in healthcare or aviation—were found to be more susceptible to AI.”

Rethinking the Role of AI in the Workplace

The new study that Quercia and his colleagues published in PNAS Nexus offers a more nuanced picture to this fraught issue.

Using the innovative approach of using AI to study itself, Quercia and his fellow authors argue that fears of mass job displacement due to AI have been overstated. While AI will undoubtedly reshape certain industries, it is more likely to augment rather than replace human labor.

It’s a finding that aligns with the Nokia Bell Labs view of Industry 5.0, in which collaborative robots – or cobots – will help people perform their jobs better and make their lives easier. New AI tools, if developed responsibly, will see machines and robots augment, rather than replace, human activities and decision-making.

In healthcare, for example, AI can assist doctors by analyzing X-rays and MRIs and performing routine tests, but it cannot replace the nuanced judgement of a medical professional. Same for an orthodontist: the AI assistant can analyze the data and perform the mundane tasks but won’t be suited for things that require more judgement and dexterity.

“No job will be fully automated. Some will be highly reconfigured but that’s about it,” said Quercia. “It’s about replacing tasks, not replacing jobs.”

Quercia, who has long argued for a more human-centric approach to AI, said that another key finding was that the professions most likely to be impacted by AI also tended to be those with large existing labor shortages. Therefore, newly automated tasks would actually help meet existing needs, such as filling in for the critical shortages in healthcare and transportation.

On the other hand, sectors with a surplus of workers, like wholesale trade and public administration, will feel more of the pain since automation will be able to do these jobs cheaply and more effectively than humans.

Those jobs that are least impacted by AI tended to be opposite sides of the spectrum, either highly skilled professions that involved interpersonal interactions or those that required basic and non-specialized skills.

The irony is that those who were typically most fearful of being replaced by AI may be those who are safest from it. In fact, the AII score that shows the lowest impact is in jobs requiring a high school diploma or less, jobs like floor sanders, farm labor contractors and quarry rock splitters.

“Most of the tasks that require manual dexterity are super hard for a robot to do and sometimes it is just too costly to replace humans with machines,” Quercia said. “In food preparation, for example, it would be overkill to put multi-million-dollar robots on the job rather than having a minimum wage worker. It’s not just what technology can do, it’s what makes economic sense.”

A New Approach to AI and Employment

The study offers crucial insights for both technology developers and policymakers, arguing that the focus should shift away from replacing workers and toward creating technologies that complement human skills as AI becomes more integrated into the economy.

Quercia and his fellow researchers therefore recommend two key strategies. The first is for developers to prioritize AI tools that enhance productivity by working alongside humans rather than aiming to automate entire professions. The second is for policymakers and business leaders to invest in training and education to ensure that the workforce is prepared to adapt to AI’s evolving capabilities.

By adopting this approach, they argue the potential downsides of AI can be mitigated while maximizing its benefits—boosting productivity and improving job satisfaction without the threat of widespread unemployment.

In that spirit, Quercia and his team have also created an Atlas of AI Risks to help the public navigate these uncharted waters. The interactive tool lets users explore over 350 real-world AI applications and classifies them by risk levels under the EU AI Act, ranking each from low to unacceptable risk.

It’s all part of a roadmap to help people adapt to a new reality, which will likely be less dystopian than some have feared but will still require preparation for a future in which humans and machines will have to learn how to work side by side.

“This is definitely a more optimistic message compared to all the news about mass unemployment,” Quercia concluded. “But the concern here is that it will require reskilling. New jobs will emerge, but the existing ones will be redefined and new education will be needed.”