If you’re like me, your first experience with generative AI bots like ChatGPT probably went something like this: You asked it to do something, it did the thing you asked it to, and you were impressed with how well it did it. Then, an existential thought crossed your mind: “Shit…can this thing do my job?”
The question of whether AI will eventually take jobs away from us meatbags is nothing new. However, following ChatGPT’s launch late last year, the speed at which AI has caught on has surprised almost everybody, even those working in the space. And far from a question to consider in the far-off (or even near-term) future, jobs are already being affected: Some layoffs this year came due to companies believing AI could replace certain roles, while other companies froze hiring for similar reasons.
So, how do you know if your job is one of the ones at risk? A recent study could give you the answer (and you might not like it).
This UK study reveals the jobs “most exposed” to AI—and what that means
Assessing the random actions of various companies and getting lost in speculation do us no good. For a substantive and thoughtful discussion on the topic, there is already traditional research ongoing into how AI will affect the job market, including this recent study out of the U.K. The study, developed by the UK’s Department for Education, estimates that 10–30 per cent of jobs are automatable with AI—which, depending on your general outlook on AI, may sound like a lot, or less than you’d expect.
The study investigated the job functions and qualifications for various sectors of the workforce, looking for whether the following ten AI applications could aid in those jobs:
- Abstract strategy games
- Real-time video games
- Image recognition
- Visual question answering
- Image generation
- Reading comprehension
- Language modeling
- Translation
- Speech recognition
- Instrumental track recognition
Depending on how relevant each of these 10 functions were to a particular role, the study generated an AI Occupational Exposure (AIOE) score for the role. The higher the score, the more “exposure” that role may have to artificial intelligence.
In the initial findings, the study determined that “professional occupations,” including sectors like finance, law, and business management, tended to be more exposed to AI. In fact, they specifically found that the finance and insurance sectors were the most exposed. Building off this discovery, it seems the more advanced the qualifications necessary for the role, the more AI exposure that role tends to have. In general, if your job requires more education and more advanced training, chances are it pairs with well AI.
The reverse is true, of course—except for security guards, interestingly enough. The study says there is such an emergence of security technology that although the role requires low education and work experience, it is more exposed to AI than other jobs of its kind.
None of this is necessarily a bad thing. As the study points out, the International Labor Organization has found most jobs are only partially exposed to AI, so the odds are decent employees in these roles will benefit from AI exposure, rather than have their jobs fully replaced by the technology.
Which jobs are most exposed to AI
Taking all this into consideration, the study breaks down the top 20 occupation types most exposed to AI, as well as the most exposed to large language modelling (LLMs). It’s a long list, including sectors like consulting, telephone sales, psychologists, legal professionals, teachers, and payroll managers.
![chart showing the 20 job categories most exposed to AI and LLMs](https://lifehacker.com/imagery/articles/01HHD0JY3SSJEG2H1A09VBTTEY/images-1.fill.size_721x750.v1702317619.png)
As stated above, the study finds that finance and insurance are the most exposed to AI of any job sector. The other most exposed sectors include information and communication; professional, scientific and technical property; public administration and defence; and education.
Just as interesting as the list of occupation types most exposed is the list of those least exposed. Many of these roles require manual labour that cannot be replicated by AI or technology in general, such as sports players, roofers, fork-lift truck drivers, painters, window cleaners, and bricklayers:
![UK study results for jobs least exposed to AI/LLMs](https://lifehacker.com/imagery/articles/01HHD0JY3SSJEG2H1A09VBTTEY/images-2.fill.size_2000x1679.v1702317619.png)
Of the “professional” occupations, the following are deemed the least exposed to AI: veterinarians, medical radiographers, dental practitioners, physiotherapists, and senior
police officers. The industries least exposed include food services; motor trades, agriculture, forestry, and fishing; transport and storage; and construction.
Will AI truly replace any jobs, according to the study?
Interestingly enough, the study is almost exclusively focused on AI exposure, rather than on jobs threatened by the technology. That said, they do have a list of 16 job types that are considered “high automation occupations,” which a pessimist could infer to mean jobs that could be one day replaced by automation.
- Authors, writers and translators
- Bank and post office clerks
- Bookkeepers, payroll managers and wages clerks
- Brokers Call and contact centre occupations
- Customer service occupations n.e.c.
- Finance officers
- Financial administrative occupations n.e.c
- Human resources administrative occupations
- Librarians
- Market research interviewers
- Other administrative occupations n.e.c.
- Pensions and insurance clerks and assistants
- Telephone salespersons
- Travel agents
- Typists and related keyboard occupations
You might notice some overlap between this list and the list of jobs most exposed to AI. That’s because the study notes that these jobs all have high AIOE scores, both for exposure to AI and LLMs.
I can’t predict the future, and neither can this study. In fact, it cautions that the results generated here are based on assumptions that may or may not prove to be true as time goes on. That said, this is a good look at the situation as it stands from a data-driven perspective. And for the record, I’m not happy to see “writers” at the top of that automation list.
Leave a Reply
You must be logged in to post a comment.