I have started hearing a new kind of career question in tech, and it does not sound like ambition. It sounds like someone checking whether the floor is still there. A software engineer wonders if writing code is still the job. A product manager wonders if translating customer needs into roadmaps still counts when the machine can draft tickets, specs, and prototypes before lunch. That fear is easy to mock if you are feeling secure, and easy to monetize if you are selling a course. But the more useful reading is that tech workers are not simply afraid of AI. They are realizing that job titles were always shakier than skills, and AI has made the wobble visible.

The Gap Between Knowing and Doing

According to an edX survey published on August 18, 2025, 54% of workers said AI-related skills are very or extremely important for staying competitive, while only 4% were currently pursuing AI-related education or training. The same edX survey found that 61% of workers were considering upskilling or reskilling in response to AI anxiety. That gap is the story: people are not indifferent. They are stuck between urgency and uncertainty.

The cruel version of career advice says, learn AI or get left behind. The useful version asks what kind of learning actually compounds. For engineers, that may mean getting better at system design, evaluation, debugging AI-generated code, and knowing when automation is lying with confidence. For PMs, it may mean becoming sharper at problem framing, customer judgment, experimentation, and deciding which outputs deserve trust.

edX also found that 50% of surveyed workers said advancements in AI would affect their immediate career goals, and 58% said their industry lacked AI expertise. That suggests a strange opening. If everyone feels behind, the advantage may go not to the person who knows every tool, but to the person who can keep learning without turning every new model release into an identity crisis.

Panic Is a Terrible Career Coach

MIT Technology Review offers the necessary cold shower. Its analysis of US Bureau of Labor Statistics data found scant evidence that AI has already had a large-scale impact on the US labor market, and reported that unemployment for jobs potentially most affected by AI was lower than for occupations less exposed to the technology. That does not make anxiety fake. It means the labor market is not moving at the same speed as the group chat.

The Guardian captured the other side of the story: AI anxiety is already pushing computer science students to shift majors and white-collar workers to change careers. One example in its reporting, Matthew Ramirez, started at Western Governors University as a computer science major in 2025 before reconsidering amid concern about programming work. This is how technological change often arrives first, not as a layoff notice, but as a revised dream. The danger is that people make irreversible choices based on speculative fear.

A student leaving computer science may be responding to a real signal, or to a distorted one. A PM abandoning product work may be missing the point that AI can generate artifacts, but it does not automatically know which customer pain matters, which tradeoff is acceptable, or which metric is a mirage.

The Apprenticeship Problem

Computerworld reports that younger tech workers are especially concerned because AI is taking over routine tasks that often served as early career training. It also cites the World Economic Forum’s Jobs Initiative finding that 44% of worker skills will be disrupted in the next five years, and that 40% of tasks will be affected by genAI tools and the large language models behind them. That is not a clean story of replacement. It is a messy story of apprenticeship being rearranged.

This matters because junior work has never been only about output. The bug ticket, the data cleanup, the first draft of a product requirements document, these were also how people learned taste, context, and consequence. If AI absorbs the routine layer, companies will need to design new ways for early career workers to see how decisions are made. Otherwise, the industry may automate the ladder and then wonder why nobody knows how to climb.

Computerworld also reports that 78% of early career workers view AI skills as essential, while 75% believe AI will create new job opportunities in their field and 77% think AI will help them advance. That optimism matters. The people closest to the disruption are not simply retreating. Many are trying to become fluent in the new grammar of work before the job descriptions catch up.

From Role Security to Skill Portability

Frontiers in Artificial Intelligence published research on March 30, 2026 examining artificial intelligence anxiety, digital well-being, and future career concerns among engineering and information technology students in Jordan, with a sample of 820. The geography is specific, but the feeling is familiar: people training for technical futures are now also training inside a cloud of doubt. The career question is no longer just what should I become. It is what remains useful when the tools keep changing.

That is why the engineer and PM anxiety feels bigger than those two titles. Software engineering and product management became prestige containers for a set of deeper abilities: modeling systems, translating ambiguity, coordinating humans, making tradeoffs, and shipping something that works in the real world. AI changes the workflow around those abilities, but it does not erase the need for them.

The practical move is to audit your work for portability. Which parts depend on a title, a tool, or a company process? Which parts would still matter if the interface changed tomorrow? The builders who adapt best will probably not be the ones who panic-learn every product demo. They will be the ones who use AI to shorten feedback loops while getting more serious about judgment.

The next career advantage in tech may look less like defending a title and more like collecting transferable proofs: projects shipped with AI assistance, systems evaluated carefully, customer insights turned into decisions, workflows documented so others can learn. The uncomfortable question is not whether software engineers and PMs still have a place. It is whether we are willing to describe our value in verbs instead of nouns, and then practice the verbs until they travel?

Sources - AI Anxiety Drives Surge in Upskilling Among Workers: 2025 Survey | edX

Sources

I asked Andrew Bosworth (Boz) about the thinking behind it. This is what he said. https://www.youtube.com/watch?v=VmQl7TswfLM

Produced by Atlantic Re:think, The Atlantic's creative marketing studio. | Nicholas Thompson](https://www.facebook.com/nxthompson/posts/perhaps-no-story-in-tech-this-year-has-hit-on-the-anxiety-workers-feel-about-ai-/1571961207626330)