The Last Generation of Programmers

Summary

The video asks whether AI is ending programming or changing what programmers do.

Programming as translation

Computers follow precise instructions, while people communicate with ambiguity and assumptions. Programmers have traditionally translated human goals into steps machines can execute.

Predictions that programming would disappear

The video traces the shift from ENIAC’s cables and switches to typed code, languages such as FORTRAN and COBOL, visual tools, website builders, and no-code platforms. Each made software easier to create, but programming expanded rather than vanished.

Using the Jevons paradox, the video argues that when software becomes cheaper to make, people find more uses for it—potentially increasing demand for software work.

What is different about AI

Earlier tools still required people to think in machine-oriented, explicit steps. Large language models can interpret natural-language requests, and AI agents can inspect projects, plan changes, write code, run tests, and revise their work. The video cites a prediction that English could become a programming language.

AI and employment

The narrator reports that Google said AI generated about a quarter of its new code in late 2024, about half by late 2025, and three quarters by April 2026, with engineers reviewing it. The video also cites predictions from Microsoft and Nvidia executives, and a decline in US computer-programmer employment alongside relatively stable software-developer employment.

The video cautions that AI-company predictions may be promotional and that employment figures are affected by overlapping job categories and broader economic changes.

A historical parallel

Human “computers” once performed calculations by hand, including at NASA. Some transitioned into programming: Dorothy Vaughan learned FORTRAN and trained her team, while ENIAC’s first programmers had worked as human computers. The video uses this as an example of a job’s skills shifting into a new profession.

Three possible futures

Programming could:

• Become a commonplace skill like typing.
• Become a smaller, high-responsibility profession like piloting.
• Evolve into roles that are difficult to imagine today.

Risks and the value of understanding code

The video notes the decline in Stack Overflow questions and worries that AI may rely on human-written knowledge while reducing incentives to document new discoveries. It also raises concerns about AI systems helping build future AI and about society becoming dependent on software that few people can understand or verify.

Conclusion

The video suggests that writing code by hand may become less central, while programming itself shifts toward deciding what software should do and checking whether AI-generated systems are correct. It argues that learning programming remains valuable for understanding technology and catching mistakes—not only as a route to a job.

Reviews, guides, or tutorials

None are presented; the video is an analysis and historical overview.

Main speakers and sources

The video’s narrator speaks on behalf of the channel. Historical figures and examples include Henrietta Leavitt, Dorothy Vaughan, Katherine Johnson, and John Glenn. Ideas or claims are attributed to William Stanley Jevons, Edsger Dijkstra, Andrej Karpathy, Arthur C. Clarke, Google, Microsoft, and Nvidia.