I am going to be bleak for a moment. You should probably sit down.
AI is going to take a lot of jobs. Not in some distant science fiction future. Soon. Maybe not your specific job in the next twelve months, but within the next decade, the employment landscape is going to look radically different. And I do not think we are ready.
This is not doomerism. This is observation. The capabilities are improving faster than anyone predicted. The economic incentives are obvious. The companies building this technology are not doing it for fun. They are doing it because automation at scale is the most valuable thing in human history.
We are living in the good years. Enjoy them while they remain.
For the last few years, the standard response to AI anxiety has been: AI will not replace jobs, it will augment them. Every technological revolution creates more jobs than it destroys. This time is no different. I used to find this comforting. I do not anymore.
Previous technological revolutions displaced workers gradually, over decades. People had time to retrain. New industries emerged to absorb the displaced. The transition was painful but manageable.
AI is not gradual. The improvement curve is steep and accelerating. GPT-5.2 makes the models from two years ago look like toys. Claude Opus 4.5 reasons through problems that would have been impossible in 2023. Gemini 3 processes multimodal inputs like it is nothing. Each generation leapfrogs the last.
We are not dealing with incremental change. We are dealing with exponential capability growth. The jobs that emerge to manage AI will be fewer than the jobs AI replaces. That is the whole point. If you needed the same number of humans, AI would not be worth the investment. The economic value of AI is that it allows you to do more with fewer people.
The factory automation fears from the 80s and 90s were real, but they mostly affected blue collar work. Robots on assembly lines. Automated warehouses. The knowledge workers felt safe. They should not feel safe anymore.
AI writes code. AI writes marketing copy. AI handles customer service. AI does legal research. AI analyses data. AI creates presentations. The things that white collar workers do, the things that required education and training and professional expertise, are increasingly things that AI does adequately.
Adequately is the key word. AI does not need to be better than the best humans. It needs to be good enough and cheaper. For a lot of tasks, it already is.
The junior roles are going first. Entry-level positions where the work is structured and the outputs are verifiable. The jobs that used to be stepping stones to senior roles are disappearing. This is going to create weird gaps in talent pipelines that we have not fully reckoned with yet.
I do not know when AI will be able to do your specific job. Neither does anyone else. The predictions range from imminent to never. The honest answer is uncertainty.
But the direction is clear. AI capabilities are increasing. Costs are decreasing. Adoption is growing. Every year, more tasks become automatable. Every year, the economic case for human labour weakens for more categories of work.
You can choose to believe this will plateau. Maybe it will. Maybe we hit a wall and current capabilities are close to the ceiling. But betting your career on a plateau that has not happened yet seems risky. The safer assumption is that AI will keep improving. The safer plan is to position yourself for a world where AI can do most of what you currently do.
I do not have a clean answer. If I did, I would be selling courses instead of writing blog posts. Some thoughts though.
Skills that involve physical presence in the real world are harder to automate. Trades, healthcare that requires hands on care, anything where being a body in a space matters.
Skills that involve genuine creativity, the kind that synthesises information in novel ways rather than recombining patterns, might retain value longer. AI is good at average. It struggles with truly original.
Understanding AI itself has value. The people who know how to use these tools effectively, who can prompt well, who can integrate AI into workflows, who can build on top of AI capabilities, will be more valuable than people who cannot.
But ultimately, some jobs are going to disappear and the people doing them will need to find something else. This has always been true of technological change. The difference is the speed and scale.
What frustrates me is the lack of serious policy conversation about this. Governments are focused on AI safety in the apocalyptic sense, which matters, but they are not focused on the economic disruption that is already starting.
We need conversations about retraining at scale. About safety nets for displaced workers. About the distribution of productivity gains. About what society looks like when human labour is worth less than it used to be.
These conversations are not happening. Or they are happening in academic papers that nobody reads. The political system is not equipped to handle a disruption this fast. By the time policy catches up, the damage will be done.
This post is bleak. I know. But I think false optimism is worse than clear-eyed pessimism.
We are in a transition period. The jobs that exist today mostly still exist. The paycheques are still coming. The skills we have are still valuable. This is good. Appreciate it.
Build skills that seem durable. Save money if you can. Develop flexibility. Pay attention to how AI is being used in your field so you are not surprised when it arrives.
And enjoy the work while it lasts. Complain about meetings and deadlines and annoying coworkers. Appreciate the fact that you are paid to use your brain. This might not be a permanent state of affairs.
AI is coming for us all. Not tomorrow. But eventually. The question is not whether the world changes. The question is whether you will be ready when it does. I am trying to be ready. I do not know if I am succeeding. But I am trying.