Sam Altman, CEO of Open AI, in Berlin in September. More than 50 per cent of entry-level positions in banking are at risk from generative AI, research suggests.Florian Gaertner/Getty Images
Sam Sivarajan is a behavioural scientist, keynote speaker and author of the book The Uncertainty E.D.G.E. – helping leaders master the tension between hesitation and haste.
Sam Altman, chief executive officer of OpenAI, has been refreshingly blunt about the long-term effects of artificial intelligence: “Jobs are definitely going to go away, full stop.” One leading report suggests that up to 200,000 jobs in global banks – jobs focused on analyst-type roles in different functions – may be lost in the next three to five years.
Other research suggests that more than 50 per cent of entry-level positions in banking are at risk from generative AI; and almost 20 per cent of first- or mid-level managers face similar threats. Similarly, a recent Stanford University study of payroll records indicated software developer jobs declined by nearly 20 per cent from late 2022 to mid-2025, as a direct result of generative AI becoming mainstream.
The proponents of AI aren’t denying this impact; rather they position it as freeing us from mundane tasks to focus on the more value-added, higher-paying tasks. Sounds good, right? Who wouldn’t want that? There’s only one problem. How does one learn to do the higher-value tasks without going through the mundane?
Thirty years ago, I was that junior investment banker at a large global company in London, England. I was working regularly till 2 a.m. building financial models, doing research, putting PowerPoint presentations together. All tasks that AI does today in the blink of an eye and at a lower initial cost than that of employing a junior investment banker. If you are a bean counter, what is not to like about that math? And if you are a junior banker, what is not to like about getting rid of this grunt work?
Sam Sivarajan: How artificial intelligence hype is drowning out investment wisdom
Sam Sivarajan: Why do smart investors make dumb decisions?
To answer that, I want to tap into nostalgia to take you back to the original Karate Kid movie. The hero of the movie, a young boy named Daniel, wanted to learn karate to defend himself against bullies. His mentor, Mr. Miyagi, promised to teach Daniel karate but first had him wax his car (“wax on, wax off”) and paint his fence (“up, down”).
Daniel did this for a week and then complained about these menial tasks. Mr. Miyagi then attacked him while yelling “Wax on, wax off. Up, down.” Daniel had to defend himself, and he did it well. Without even realizing it, those menial tasks had built the muscle memory in him to deal with more complex challenges – such as defending himself.
I learned a similar lesson. Those mundane tasks as a junior banker were teaching me pattern recognition. They were teaching me about solving problems. They were teaching me judgment. When you build 50 financial models, you start seeing what makes a deal work and what questions to ask. When you research comparable companies for 30 deals, you develop an instinct for valuation and deal parameters.
By the time I was running deals, I wasn’t relying solely on analysis – something AI can do today – I was relying on judgment built from thousands of hours of repetition doing menial tasks.
In a world of technology and AI, what are the skills that humans will need to succeed? In my view, they are judgment, imagination and emotional connection. These are not skills we are born with; they are forged in the crucible of experience. Hard-won experience.
If AI is removing this training ground, business leaders may save a few bucks today but might very well be creating a much larger structural problem in the long term.
By eliminating junior bankers from the roster, how will they replace the senior bankers when they retire or get poached? By eliminating junior programmers developing basic code today, how will they replace the senior developers who we depend on for high-end applications?
These junior roles provide the training, the experience, the space to make relatively inconsequential mistakes and learn from them under supervision. By eliminating these roles, how do we prepare future leaders to step in? Can they step in without this grounding and training? More importantly, would we want them to step in?
In his book, When More Is Not Better, Roger Martin argues that this obsession with efficiency has made economic systems fragile. He says that this approach intensifies risks, marginalizes essential workers and reduces the system’s ability to adapt to shocks. To that, I would add that it removes the building blocks that prepare the next generation to contribute and lead in meaningful ways. Short-term focus on gains may very well lead to longer-term pain in terms of jobs, stability, community and a resilient country.
For long-term value investors, this might mean looking beyond quarterly results to see how companies are preparing the next generation of leaders.
As Marshall McLuhan observed: “We become what we behold. We shape our tools, and thereafter our tools shape us.” Yes, build better tools. Just don’t lose the skills that built them.