Will AI take my job? A Google-certified trainer told me it would. Twenty-two months later my job title is unchanged — but ninety percent of what I do is different. Why I am not afraid, and why most fear is misplaced.
Will AI take my job?
'Your profession will cease to exist.'
One of the first Google-certified trainers in Slovakia told me this at a tech summit in June 2024. I went quiet. I kept that sentence.
22 months later: my job title has not changed. Ninety percent of my work has.
Not yet. But the job as it was is already gone.
A Czech software firm quoted more than €50,000 for a simple customer-facing tool. With Claude Code, I built the same tool in one hour and fifty-five minutes. This is what that gap reveals about software pricing, AI agents, and the next year.
After 150 practical interviews for data and marketing roles, I noticed a pattern — candidates with AI access consistently underperform those from three years ago who had none. The AI productivity paradox is not theoretical — I see it every week.
3 years ago, 1 in 5 interview candidates scored above 6/10 on my analytical tasks.
Today, with access to the most powerful AI models ever built, I struggle to find anyone above 5.
150 interviews. The pattern is clear.
AI doesn't know when it's wrong. A correct answer and a fabricated one look exactly the same — to you and to the model.
Companies connect these models to emails, invoices, and personal data.
What happens when it makes a decision that costs real money?
AI models hide their system prompts, training data, confidence levels and alignment decisions from you. Eight layers of concealment — from hallucinations to corporate RLHF instructions — and what it means for businesses deploying autonomous AI agents in critical infrastructure.
