Before we ask how many tasks a machine can perform, ask what happens to the people who perform them.
That is where I would begin the workplace conversation about AI. Too often, the discussion moves quickly from a product demonstration to a prediction about whole careers. A person becomes a cost line. A profession becomes a collection of steps. The changes may be real, but the human consequences deserve more care than that.
The OECD’s Employment Outlook 2023 examined AI and work, including surveys in manufacturing and finance across seven countries. It described potential benefits alongside concerns about privacy, bias and work intensity. This is evidence from particular settings and a particular period, not a forecast of what will happen to every worker in 2026.
I take from it the need to examine actual working conditions. Does a tool remove repetitive effort, or create new pressure to do more? Does it help someone exercise their expertise, or quietly narrow their discretion? The same technology can be introduced under very different management choices. Those choices need attention.
Loading reader polls…
For smaller organisations, I can see the attraction of better assistance. Preparing a first draft, arranging information or reducing administrative friction could make room for work that has been neglected. But the saved time should have a purpose. Otherwise efficiency can become a demand that everybody run faster indefinitely.
My preference would be to bring employees into the discussion early. Ask which tasks frustrate them. Listen to where their judgement matters. Test the tool with the people expected to use it. A system selected without that knowledge may look impressive in a presentation and fit badly into the daily work.
Training should also be treated as work, not an unpaid personal obligation. People need time to learn, make mistakes safely and understand when to challenge an output. Telling a team to become adaptable while giving them no space to adapt is an unreasonable way to manage change.
There must be a route to question automated assessments as well. If AI contributes to a decision affecting a person, someone should be responsible for explaining the process and dealing with a concern. The tool should not become a convenient place to hide management responsibility.
For students preparing for employment, I would encourage curiosity about AI alongside the capacities that make collaboration possible: clear communication, careful reasoning, domain knowledge and reliability. Knowing a tool can help. Knowing why a task matters and whether the result is sound can help a great deal more.
We should resist easy promises in either direction. Nobody owes us certainty about an unsettled future. Institutions do owe people honesty about changes they are making now. What is being tested? What support is available? How will the effects on workers be evaluated? Those are questions we can answer concretely.
I want progress that enlarges opportunity and gives people a fair chance to participate. That will take more than installing software. It will take attention to dignity, skill and the distribution of benefits. Let the machine change the task where it genuinely helps. Keep the person in the conversation about what happens next.
Evidence and reading
jai.a.sam
jai.a.sam is the founder of Students Herald and writes about international education, student opportunity and sector development. His columns explore the people, ideas and decisions shaping how students learn, travel and build their futures.
