AI 对 IT 各工种的冲击¶
状态: ✅ 已完成
创建日期: 2026-03-23 最后更新: 2026-03-23
LinkedIn Post¶
I ran program management across 8 IT roles for 10 years. I thought I knew which ones AI would hit hardest.
I was wrong about most of them.
QA Engineering: The 1,359 contingency plans we ran before Singles' Day used to take months of manual prep. AI now generates 80% of the test scripts in days. The QA team didn't shrink. It got more focused. Script writers left. Failure-mode thinkers stayed.
Technical Documentation: Everyone assumed this was the first casualty. AI writes a first draft faster than any human. But the whole job is now accuracy review: does this doc match what the code actually does? That's harder than writing. Docs people who can read code and catch AI hallucinations are in more demand than before.
DBA: I've watched AI cut query optimization time by 90%. I've also watched a DBA stand between 430 engineers and a database meltdown at 2am. The optimization part, AI's doing it. The "I know this table, I know what breaks it, and I know not to do what looks obvious right now" part, that's not in any model.
TPM/PM, my own role: AI replaced the dashboard work I used to do at midnight. It didn't replace the midnight. When two teams are pointing at each other after a payment failure, AI gives me better data. It doesn't tell me who's wrong, what matters, or what to promise the business.
The jobs that changed most aren't the ones that got eliminated. They're the ones where the entry-level work disappeared and left only the parts that required judgment.
If you're a senior engineer or PM in Vancouver or Seattle navigating this, where are you landing?
TPM #AgenticAI #Vancouver¶
X Standalone Post¶
8 IT roles. 10 years. AI hit them all differently.
QA script writers left. Failure-mode thinkers stayed.
Docs became accuracy review, not writing.
The DBA who knows what breaks at 2am is irreplaceable.
Where's your role landing?
TPM #AgenticAI¶
发布记录¶
- LinkedIn: https://www.linkedin.com/feed/update/urn
activity:7441930902932406272/ （2026-03-23 12:38pm PT，192 impressions in ~1h） - X: https://x.com/XINDR365/status/2036166937693265968 （standalone post + Reply 放 LinkedIn URL）
- 配图:
10-images/linkedin_post_ai_impact_roles_20260323.png（1400×750，8 工种 AI 冲击对比信息图）
同步发布的评论¶
| 目标帖子 | 作者 | 评论内容 |
|---|---|---|
| Load Balancer vs API Gateway | Alex Xu (ByteByteGo) | We had someone put auth logic in the load balancer once. Took a full sprint to untangle. "They're not competing tools" — wish I'd had that line on a slide back then. |
| 77 Cold Email Openers | Nick Palasz (Slyleadz) | AI writes most cold outreach now. The trigger-based and observation openers on your list are exactly what still works — they prove a human actually looked. |