Commentary: AI's growing impact on jobs and the coming talent void

"This study on the jobs affected by AI is the latest in a string of similar studies from other companies, including OpenAI, Anthropic and others. All try to forecast which jobs are going to be most heavily impacted by advances in generative AI.

"Consistent with other studies, the safest jobs are physical in nature, like a nurse or a mechanic, whereas the jobs most at risk are knowledge and interaction workers, such as sales assistants and interpreters.

"The impact on jobs is, at the moment, a bit one-dimensional. We look at what a job currently entails, what an advanced AI might be able to do within that context and then try to predict the impact.

"These studies don’t, indeed can’t, take into account how job types will change and adapt to the availability of AI. A simple sales call may be replaceable by AI, but a complex sales activity, which requires a deep understanding of the individual, their responses, and their psychology, where a salesperson is aided by an AI, will not. Today’s safe job may be tomorrow’s lost profession, and vice versa.

"Take education as an example. For years, we’ve moved towards a model of learning by rote, with exams that assess what you’ve memorised. This activity is arguably better performed by AI, creating personal mentors that can be your learning buddy, answering the silliest of questions, explaining topics in ways to suit your own learning styles, and assessing your strengths and weaknesses. Is this a recipe for AI-only universities? I’d suggest not – the value of learning in a community of peers, learning life lessons, hearing from the great minds of the age, all from a human perspective, won’t die away, but it will have to change.

"Although well-educated knowledge workers seem to be at high risk of AI displacement, they are also the most likely to adopt and adapt to the use of AI, creating new job profiles and skills which don’t currently exist.

"The Microsoft analysis also doesn’t look at the importance of experience and skill. We’re seeing a number of AI rollouts being used to automate certain roles, and they tend to focus on the low-experience admin or basic knowledge worker. Graduate jobs are arguably being more impacted by advances in AI as companies seek to take an ‘AI dividend’, an immediate cost saving gained by not employing a graduate. The problem is that this creates a talent void, which will be experienced in future years as the next generation of experts and leaders will not have been trained or given the chance to gain experience. How do you get to be an editor if you’ve never had to write? How do you get to be a CFO if you’ve not accomplished basic accountancy? Short-term savings will create a long-term problem.

"There is a more insidious problem, that of progressive de-skilling. As AIs get better at knowledge aggregation and reporting, people will be less inclined to challenge a machine’s outputs, eventually coming to rely on the ‘thinking’ done by the machine and losing the ability to challenge or critique.

"There is also a problem of what might be termed ‘cultural imperialism’ due to AI platforms being used to analyse, report, and even teach, based on a Western-centric view of the world, training AIs on biased datasets designed by a small section of elite Silicon Valley companies. Views, cultures, languages, histories, etc, that aren’t regarded as a priority will not gain the same prominence in the machine’s outputs as we might like. As such views become woven into society, we run the danger of becoming a monoculture, a society with homogenised views that may not align with your country or culture’s values."