AI Killed The Best Job of the 21st Century

Summary

The video argues that programming is not dead, but that companies may be underestimating the costs and risks of adopting AI. The narrator, a software engineer, groups these concerns into three “bills”:

1. The cost of AI tools

Coding assistants charge for tokens, and autonomous tools can consume large quantities while reading, writing, and repeatedly testing code. The video cites several examples:

• Uber reportedly exceeded its annual AI-coding budget in four months.
• A Microsoft engineer reportedly spent $28,000 on AI in 28 days.
• Gartner predicts that AI coding costs could exceed the average developer’s salary by 2028.

The video suggests that falling token prices do not necessarily mean lower overall costs if usage expands.

2. Technical debt and security

Drawing on GitClear’s analysis of more than 600 million code changes, the video says copy-pasting increased while refactoring fell—from about 21% of modified code in 2022 to below 4% in 2026.

It also cites Veracode testing that found roughly 45% of AI-generated code had a known security flaw when models were not explicitly prompted to prioritize security. The narrator’s concern is that AI can produce code quickly while leaving more cleanup and review work for later.

3. Fewer opportunities for junior developers

The video describes research finding that employment among 22–25-year-olds in AI-exposed jobs was about 19% lower than expected, while more experienced workers in those fields were doing better. It argues that companies may be reducing junior hiring because AI handles many entry-level tasks.

This could weaken the traditional path through which developers learn—from mistakes, code reviews, and gradually harder assignments—leaving fewer people prepared to maintain AI-generated systems in the future.

Interpreting alarming headlines

The video cautions against taking alarming headlines at face value. It says the headline that US “computer programmer” numbers had reached their lowest point since 1980 used a narrow occupational category: many people commonly called programmers are counted as software developers instead.

It also notes that some companies may be “AI washing”—attributing layoffs to AI even when other factors contributed.

Advice for students and aspiring programmers

The narrator advises using AI as a learning aid rather than simply accepting its answers. Students should ask why AI produced particular code, understand what needs to be built, and verify that the result is correct.

The narrator expects the ability to understand and review code to remain important, but acknowledges uncertainty about whether future AI will become capable of fixing its own errors.

Reviews, guides, or tutorials

No product review or step-by-step tutorial is presented. The video offers career guidance on learning programming and using AI tools.

Main speakers and sources

• Main speaker: The video’s narrator, who identifies as a software engineer with a master’s degree.
• Sources and organizations discussed: Sam Altman and OpenAI; Anthropic and its research; GitClear; Veracode; Stanford researchers; South Korean developer researchers; Gartner; and executives or company communications from Amazon, Uber, and Microsoft.