Hacker News

New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Amazon Surprises by Actively Rehiring Former Employees (asiae.co.kr)
    —discuss
  2. Show HN: Koi.rest – watch some fish and regain your balance (koi.rest)
    —discuss
  3. US Mobile CEO on Reddit: 1.1M customers, $1B+ on Warp, 5G SA, QCI 6, satellite (reddit.com)
    —discuss
  4. Escaping Space: Part I (perplexity.ai)
    —discuss
  5. Vertical Farming: Reimagining Agriculture in an Urban Age (worldsensorium.com)
    —discuss
  6. Remembering the Future: Blade Runner (blue-continuum.com)
    —discuss
  7. Dutch designer made DE9: Closer to the Edit into a playable web-based instrument (creativeboom.com)
    1comments
  8. Becoming more than a meat proxy (codeplusconduct.substack.com)
    —discuss
  9. Meta's Muse page can't even pass basic privacy requirements (getprivisy.com)
    —discuss
  10. New England Aquarium treating elderly penguin, 29, with acupuncture (neaq.org)
    —discuss
  11. Should we stop giving cyber threats fancy names? (techzine.eu)
    —discuss
  12. Why Buran Had Four Computers, Not Three (zatona.dev)
    —discuss
  13. Show HN: Public Browser MCP – +48% speed -33% token/session vs. AgentBrowser (github.com/silbercue)
    —discuss
  14. Education Consulting Industry Sucks (rivernova.vercel.app)
    —discuss
  15. Show HN: I rebuilt my company's old fire-station office as a Quake III map (sajarin.com)
    —discuss
  16. The New STOC Rules for the AI Era (computationalcomplexity.org)
    —discuss
  17. Show HN: Turn Your Voice into Art (reotoi.vercel.app)
    —discuss
  18. Empirical results fine-tuning π0.5 on a real manufacturing task (dream-machines.eu)
    —discuss
  19. IR (2022) (github.com/dstogov)
    —discuss
  20. Meta and the new VR glasses? Will Apple respond?
    1comments
  21. China's new travel rules unsettle tech giants and talent (dw.com)
    3comments
  22. Hail is a multibillion-dollar problem – and could get worse (hcn.org)
    —discuss
  23. Vibe Coding Production Kit – a production workflow for AI coding agents (github.com/moeeryani)
    1comments
  24. Show HN: PipeFitter – Quick, Browser-Based, Parametric Pipe Connectors (mankyd.github.io)
    —discuss
  25. Show HN: Free WCAG accessibility checker – no signup, results in ~2 minutes (wcagrepair.com)
    —discuss
  26. AI Help for Biological or Chemical Weapons (science.org)
    —discuss
  27. Show HN: Canary (YC) – Independent verification for AI code (runcanary.ai)
    —discuss
  28. Dampen camera motion and track the leftover, in WASM (residual.dog)
    —discuss
  29. International observers to investigate Swedish election fraud (tv4.se)
    1comments
  30. SkillOpt: Training Loop for Agent Skills (microsoft.github.io)
    2comments

Ask HN: How would you know if you have learned something?

2 pointsby 1h ago
0 comments
I have been thinking about this while working on building my own AI-based tutoring platform Socratix.

AI made learning really easy, and that was the problem that made me worried. You can get an explanation anytime, nod, feel that you have understood it all, and in two days time fail to reconstruct it.

On top of that there are hallucinations, inconsistencies, and information overload. I got really tired of all this.

Thus, I decided to build my MVP based on one constraint only: nothing is "done" until it is defended.

It is going to be a dependency-based curriculum, AI-validated lessons, sandboxed coding challenges, and at some point a short oral defense to get to the next topic.

It is early and I might change the name. But I am sure that I am doing some things wrong that I would not do otherwise.

What bothers me the most is the oral defense. Is it a good indicator of your understanding or just another hard test that looks fancy?

Really appreciate your comments or discussions about the MVP if you are interested.

A quiet thread, for now.Start the conversation on HN ↗