AI has written all code for a year, but is not taking any tech jobs yet. Why?
retained text
I originally thought there was going to be an AI job apocalypse when AI started doing all my code. But it has not happened yet. My prediction did not turn out to be true, and I am trying to understand why while being neutral and unbiased. Exhibit A: Software job postings have gained modestly since 2024 but still down from all time highs post pandemic. Interest rates killed tech jobs, not AI \[https://fred.stlouisfed.org/series/IHLIDXUSTPSOFTDEVE\](https://fred.stlouisfed.org/series/IHLIDXUSTPSOFTDEVE) Exhibit B: All big tech companies are nearly back to their 2022 headcounts. Sure, zero job growth over 4 years isn't great, but not an AI apocalypse, especially when you realize the first rounds of job cuts were driven by rates \[macrotrends.net/stocks/charts/MSFT/microsoft/number-of-employees\](http://macrotrends.net/stocks/charts/MSFT/microsoft/number-of-employees) \[macrotrends.net/stocks/charts/GOOG/alphabet/number-of-employees\](http://macrotrends.net/stocks/charts/GOOG/alphabet/number-of-employees) \[macrotrends.net/stocks/charts/META/meta-platforms/number-of-employees\](http://macrotrends.net/stocks/charts/META/meta-platforms/number-of-employees) \[https://www.macrotrends.net/stocks/charts/aapl/apple/number-of-employees\](https://www.macrotrends.net/stocks/charts/aapl/apple/number-of-employees) Exhibit C: \[https://siepr.stanford.edu/publications/policy-brief/what-really-happening-jobs-separating-ai-hype-reality\](https://siepr.stanford.edu/publications/policy-brief/what-really-happening-jobs-separating-ai-hype-reality) AI has not meaningfully disrupted AI exposed jobs. Entry level postings have gone down, but they were already following that trend when rates shot up in 22. My theories: 1. Writing code was rarely, if ever, the bottleneck in software development. Planning, meeting with end users, reviewing code, actually adding value to the company was the hard part 2. AI has induced demand by making previously out-of-reach projects doable and has created entirely new projects. Most of the projects I have worked on this year would have been impossible a year ago 3. Companies are too scared to torch their headcount. There is a lot of valuable domain knowledge in the Amazon, Google, etc. workforce 4. AI has meaningfully added time and complication to code reviews. Gargantuan 5000 line PRs of AI slop are very difficult to review 5. Cybersecurity. For AI to meaningfully take jobs, it will need to be able to write to prod, edit files, create database tables, and so forth. Most of these permissions are gated to experienced developers
document fetched 2026-10-20 00:19 · canonical https://reddit.com/r/artificial/comments/1xa7q6o/ai_has_written_all_code_for_a_year_but_is_not