One question currently permeates our social conscience: to what extent our jobs are at risk. Beneath predictions that AI will transform the workplace and upend education lies the more unsettling concern that machines might render human effort itself redundant. For students, the stakes are immediate. We are spending years acquiring the skills and qualifications meant to secure our futures, just as the value of both is being called into question. Is AI coming for our careers, or quietly dismantling the very ladder we are trying to climb?
The danger may not be that AI eliminates entire professions overnight, but that it quietly erodes the first rung of the career ladder. In its coverage of AI at work, the Financial Times highlighted how junior bankers risk losing opportunities to develop professional skills when AI generates presentations that once required them to learn the craft themselves. Similar pressures could extend across other professions, from legal research and financial analysis to journalism, marketing and engineering, as employers increasingly find that tasks once assigned to junior staff can be completed by AI.
Meanwhile, the competition to secure that first foothold is already fierce; applications for internships are constantly rising, even as available places fall. This doesn’t mean that AI is necessarily responsible for the shrinking supply of opportunities. It does, however, illustrate the uncertainty facing students entering an increasingly competitive market. The grunt work of a first job may not be glamorous, but it is the soil in which professional judgement takes root. If that ground is paved over by automation, tomorrow’s experts may find the ladder to their careers missing its bottom rungs.
After years of Turnitin assignments and hours spent wrestling with reading lists, students might reasonably expect a degree to be evidence of what they have learnt. Yet AI can now produce a passable essay or condense a week’s reading into a handful of bullet points, sparing students some of the more threadbare aspects of university life. The difficulty is that, now, the finished product can be hard to distinguish from the understanding it is supposed to demonstrate.
The same problem is following graduates into job interviews. The Guardian recently reported that employers are increasingly concerned about candidates using AI to present abilities they may not actually possess. That is not to say, however, that a degree is worthless simply because AI can imitate its outputs. Rather, it exposes the difficulty universities face in distinguishing genuine student understanding from convincing machine imitation. The same uncertainty that troubles employers when assessing graduates also challenges universities to reconsider how they assess students. Future assessments might have to place greater weight on independently-completed work under exam conditions and oral evaluation, to help ensure that qualifications reflect individual student ability. After all, a degree ought to demonstrate intelligence, not merely access to it.
Our own student experiences may offer an insight into an issue that can otherwise feel rather abstract. As a Classics student, I can understand that translating a Latin poem, for instance, involves more than arriving at an English equivalent. Interpreting the nuances of the original and understanding how its imagery works within a wider literary tradition are key parts of the process. One fourth-year student I spoke to admitted that he had “a gut reaction to ask AI in order to then choose a way to think” about a question posed to him concerning independence of thought.
He also drew a distinction between gleaning information from AI and from different secondary sources: “AI is like giving someone the Lego bricks and instructions and having them return the finished model. Reading, by contrast, means getting different parts of the instructions from different sources and putting the model together yourself.” The point is not that thought can never be entirely independent of others, but that knowledge only becomes our own when we have worked out what to make of it. The danger is that, in letting AI build the model, we end up a few bricks shy of a load.
There is, of course, a perfectly reasonable case for letting machines do things that machines can do better. Few students will mourn the loss of an unnecessarily tedious assignment, and fewer still would volunteer to spend an afternoon formatting a banker’s presentation. But efficiency is a peculiar measure of progress when the supposedly wasted effort is how we acquire expertise. If we can bypass the climb and have the answers readily supplied, what is left of the achievement when the effort behind it can be outsourced? Perhaps the ultimate luxury in an AI-powered world will not be having the answer, but having had the opportunity to work it out for ourselves.
Image by Igor Omilaev on Unsplash.
