AI & Future
AI is reshaping work, but the story is more nuanced than the headlines suggest. Here is a calm, balanced look at what is shifting and how to stay adaptable.
AI is automating specific tasks — drafting emails, summarizing reports, writing boilerplate code, producing first-draft translations — much faster than it is deleting whole jobs. So the useful question for your own career is not "will my job disappear," but "which of my weekly tasks are exposed to a large language model, and which parts of my work stay stubbornly human."
When people say "AI" in a work context today, they almost always mean generative AI: large language models (LLMs) such as OpenAI's GPT-4o, Anthropic's Claude, and Google's Gemini, plus image tools like Midjourney, DALL-E, and Adobe Firefly. This wave differs from the automation that came before. Older software automated structured, rule-based work — a spreadsheet formula, a payroll run, a barcode scanner. Generative AI handles unstructured language and images: it can draft a memo, answer a support ticket, or summarize a 40-page contract — the exact tasks that used to require a person precisely because they involved messy human text.
That distinction matters because it changes who feels the effect. Earlier automation hit factory floors and routine back offices. This wave lands on knowledge work — writing, coding, analysis, customer service, design — jobs that were long assumed to be safe.
A 2023 study by researchers at OpenAI and the University of Pennsylvania ("GPTs are GPTs") estimated that roughly 80% of US workers could see at least 10% of their tasks affected by LLMs, and about 19% could see 50% or more of their tasks affected. Read carefully, that is a measure of task exposure, not job loss — and the gap between those two things is the whole story.
Consider concrete roles:
In each case the job is a bundle of tasks, and AI unbundles it. The routine, repeatable portion shrinks; the judgment, relationship, and accountability portion grows in relative weight.
The evidence so far points to augmentation more than wholesale replacement, and the numbers are specific:
That second finding is the important one: generative AI often compresses the gap between novices and experts. It raises the floor faster than the ceiling, which quietly reshapes what "entry-level" work is worth.
But replacement is real too, and it is honest to name it. Klarna said in 2024 that its OpenAI-powered assistant was doing the chat work equivalent of roughly 700 full-time agents. IBM's CEO said in 2023 the company expected to slow or pause hiring for around 7,800 back-office roles it believed AI could take on over several years. Both are real, and both cut against a purely rosy story.
Here is the part most people get wrong, and it is not hypothetical. In 2025, Klarna publicly walked back some of its aggressive automation, with its CEO admitting the company had leaned too far on AI and let customer-service quality slip — then moving to put humans back in the loop. That arc — over-automate, hit a quality wall, rehire for judgment — is now common enough to be predictable.
The lesson for individuals: the safest work is precisely the work that fails loudly when it is wrong. A mildly clumsy summary is tolerable; a wrong medical read, a botched legal filing, or a tone-deaf reply to a furious customer is not. Accountability does not automate well, because someone still has to own the outcome.
Since the specific tools change every few months, bet on capabilities that outlast any single model.
Learn to write a precise prompt — then, more importantly, to check the result. LLMs "hallucinate," producing fluent, confident, wrong answers. The person who can catch the wrong 5% is worth far more than the person who pastes output unread. This verification skill is fast becoming its own recognized competency.
Because AI raises the floor, surface-level competence gets cheaper. What holds value is knowing why an answer is wrong, what a client actually needs, and which trade-off matters — the context a model does not have.
Communication, trust, negotiation, ethics, and handling genuine ambiguity remain where current AI is weakest. Roles built on these — nursing, skilled trades, complex sales, leadership — shift slowly, task by task.
Skip the abstract worry and do this instead:
Most people do the opposite: they either ignore AI entirely or assume it can do everything. Both are wrong for the same reason — they never tested it against their actual work.
For most roles, no — but it will likely change the mix of tasks within your job. The higher the share of routine, predictable, text- or data-based work you do, the sooner and harder you will feel it, and the more worthwhile it is to build the judgment- and relationship-heavy parts that automate poorly.
Work that combines physical presence, human trust, and high-stakes judgment holds up best: skilled trades such as electricians and plumbers, nursing and hands-on care, and roles where being wrong is expensive and someone must be accountable. These evolve gradually, task by task, rather than vanishing.
Yes — but treat the specific tool as disposable and the skill as durable. The valuable ability is directing any AI tool well and verifying its output, not memorizing one product's menus, because the leading model will almost certainly be different a year from now.
Treat precise forecasts ("X% of jobs gone by year Y") with skepticism; they are estimates of task exposure dressed up as certainty, and past technology predictions have often missed badly. What is reliable is the direction — tasks shift, new work appears unevenly — not the exact numbers or timing.
Nova spent years as the unofficial tech-support person for everyone she knew before founding Clixvia to do it at scale. She believes technology should serve people, not baffle them, and writes clear, calm guides that treat readers as smart adults who simply weren't handed a manual. She has a low tolerance for jargon and a soft spot for a well-labeled settings menu.
Keep reading
AI news moves fast and most of it is noise. Here is a calm, jargon-free system for staying informed about what matters without burning out on every headline.
Priya Nadar
Recommendation systems shape what you watch, buy, and read every day. Here is a clear, jargon-free look at how they work and how to stay in control.
Priya Nadar