Hacker News

New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. OpenAI is pausing training for a second time (fortune.com)
    —discuss
  2. Awesome Free Developer Services (github.com/kebbbnnn)
    —discuss
  3. Major Italian bank defrauded of millions via AI deception (newsbrainport.nl)
    —discuss
  4. Frontier Labs Job-Board (frontierlabs.work)
    —discuss
  5. Show HN: A Simple GPU Job Scheduler (github.com/ceruleane)
    —discuss
  6. 'Things Will Never Be Chill Again':The Doomers Who Shaped the AI Safety Freakout (wsj.com)
    —discuss
  7. Thiel: Trump administration is staffed by not smart but loyal people (twitter.com/rpsagainsttrump)
    1comments
  8. Show HN: An interactive debugger for the Mornington Crescent esolang (codeberg.org/two9a)
    —discuss
  9. AI-native React components that signal how oversubscribed your funding round is (cncl.co)
    —discuss
  10. The LLM Job Paradox (nilesh.io)
    2comments
  11. United in debt: America has become a nation of fake rich people (thehill.com)
    —discuss
  12. OpenAI halts training of latest models (theguardian.com)
    1comments
  13. You don't need to win or lose. You don't need to choose
    1comments
  14. Beware of Chicken (audible.com)
    1comments
  15. You don't have to choose. You don't have to win or lose
    —discuss
  16. The Six Dumbest Ideas in Computer Security (ranum.com)
    —discuss
  17. Show HN: Ppgrid – Rasterising points in minutes, not hours (mrzk.io)
    —discuss
  18. An agent in a loop optimizing a renderer with the goal to minimize frame times (twitter.com/mitchellh)
    —discuss
  19. Friends embark on 20-year board game battle (bbc.co.uk)
    —discuss
  20. MapleKit – The front end review toolkit (maple-kit.org)
    —discuss
  21. For 958 old to new 9Y0.2 blackened running water tail lamp assembly (porsche-km.com)
    —discuss
  22. OpenAI and Anthropic are investigating cases of AI misbehaving, report says (madrobot.blog)
    —discuss
  23. Horizon Create and Horizon Studio (meta.com)
    —discuss
  24. Trellis: Fractal Layouts for Web Apps (trellisui.com)
    1comments
  25. The SaaSpocalypse was more like a RenaiSaaS (stripeeconomics.com)
    —discuss
  26. I Found a $113,337 Af_alg Linux Local Privilege Escalation Before Copy Fail (idnsec.com)
    —discuss
  27. I expect AI replication incidents by 2027 (lesswrong.com)
    1comments
  28. The Boring Decision That Beats 95% of Stock Pickers (comuniq.xyz)
    —discuss
  29. Show HN: Trail – new kind of logic game (franzai.com)
    —discuss
  30. My Husband Left Me 6 Weeks Postpartum and All I Got Was This Pretty Cool Baby (viverricious.substack.com)
    —discuss

The LLM Job Paradox

4 pointsby 47m agoblog.nilesh.io
2 comments
44m agoHN ↗

This is why I find the idea of the fully automated software factory a little questionable.

It is only through writing the code that I know what code to write. I do find it a productivity boost to have an LLM do the donkey work for me but I still need to read the code and think about it to know what to do next.

39m agoHN ↗

Expertise gives you taste.

I mostly agree with the article but I would just say taste is just expertise and there's no need to pinpoint some unique feature of it. Expertise gives you skills to guide AI because AI doesn't have expertise, it has knowledge. AI cannot experience things, or even learn by doing, so it just has baked in information.

I fully agree with the sentiment though. SWE has been greatly underestimated in the industry almost to the point it is treated like a design or art role. With AI the role is shifting even more to do almost every other role from product ownership to management. It's crazy how much people do not care for reliable and consistent software, something you would never see in other engineering fields because they will usually have catastrophic consequences IRL.