September 23, 2026 · by David Gilbert · 4 min read · Tech & AI
I get asked fairly often, usually by someone either starting out themselves or worried about a younger relative's career prospects, what actually happens to entry-level tech jobs as AI tools keep getting more capable. I don't think the honest answer is either the comforting version or the doom-laden version that tend to dominate this conversation online. Here's my actual, considered take.
What's Genuinely Changing, Honestly
A meaningful share of traditional entry-level tasks — basic, repetitive code generation, simple data entry and categorisation, first-draft routine documentation — are exactly the kind of well-defined, repetitive work these tools are currently good at, and getting better at quickly. Pretending this isn't happening, to make anyone feel better in the short term, doesn't serve anyone well in the slightly longer run.
What I Don't Think Is Happening
I don't think this means entry-level tech roles disappear entirely, full stop. I think it means the actual content of those roles shifts meaningfully, the same general pattern that's repeated across plenty of previous waves of automation throughout history — the specific tasks change considerably, the underlying need for capable people in the field generally persists, sometimes in a genuinely different, less immediately obvious shape than before.
What I Think Actually Matters More Now for Someone Starting Out
Judgement, genuine context, and the ability to evaluate and meaningfully direct AI-assisted output matter considerably more relative to pure raw technical execution than they used to. Someone who can review, catch errors in, and meaningfully improve AI-generated work is currently in a genuinely stronger position than someone who can only produce similar work entirely manually but can't evaluate either version critically.
Why This Is Genuinely Harder for Someone Just Starting Out
Judgement has always, traditionally, been built through years of doing the more basic, repetitive work first and slowly developing real intuition through that accumulated practice. If AI tools increasingly handle a meaningful share of that traditional, repetitive foundational work, the path to building genuine judgement looks meaningfully different than it did for people, myself included, who built it the slower, more traditional way over years.
What I'd Actually Tell Someone Starting Out Right Now
Use the tools deliberately and openly, rather than avoiding them out of principle or fear, but make a genuine, deliberate point of understanding why a given output is correct or incorrect, not just whether it happens to work in this one specific instance. That deliberate habit, actively practised, builds judgement even when the AI tool is doing a meaningful share of the actual mechanical execution underneath it.
Why I'm Not in the Doom Camp on This
Every previous wave of significant automation eventually created new categories of genuinely valuable work that didn't exist in quite that form before, even while it real, genuinely disrupted some existing roles along the way. I have no special, reliable ability to predict exactly what those new categories will specifically look like for tech work this time around. I'd bet, with real confidence, that they'll exist in some recognisable form, the same way they reliably have in every previous wave of major technological disruption I can think of.
Why I'm Also Not in the Pure Comfort Camp
Telling someone "don't worry, it'll all be fine, exactly the way it's always been" isn't honest either, and it doesn't actually help anyone prepare properly for a genuinely real transition that's already clearly underway. The honest, more useful answer sits somewhere specifically in between blind comfort and blind panic: real change is genuinely happening, the field doesn't disappear, and the actual skills worth deliberately building now are shifting in a way that's worth taking seriously and adjusting for.
The Practical Advice I'd Actually Leave Someone With
Get genuinely comfortable with the tools rather than avoiding them, build judgement deliberately rather than assuming it for granted the way earlier generations could, and stay adaptable rather than assuming any specific tool or skill is permanently safe and immune from further change. That's honestly the same advice I'd give for thriving through any major period of technological change. This one just happens to be the specific one currently in front of us.