- Author
- Posted by zaikunzhang, with replies from many commenters
- Published
- 5 April 2026
- Kind
- Hacker News discussion thread, 991 points
- Original
- news.ycombinator.com/item?id=47647788
- Subject
- Review, taste and expertise
- Themes
- Skill under automation, Getting good
What it is
This is a discussion of an ergosphere.blog post. The post imagines Alice and Bob producing equal papers while only Alice learns. Commenters, many of them software developers, debate whether that matters.
What it says
- Commenter stavros argues Bob got the same useful result as Alice with an agent.
- Several replies say the program exists to produce scientists, not results.
- Several commenters reject the calculator analogy because students learn arithmetic by hand first.
- Commenter sd9 expects the market to stop valuing the skills Bob skips.
- Commenter klabb3 warns the AI gives plausible answers Bob cannot judge as right or wrong.
- Commenter mtrifonov says the drift comes from the agent setup, not from AI itself.
What evidence it gives
The thread offers opinion, analogy and personal anecdote. Commenters mention calculators, cooking, assembly language and their own jobs. A few cite figures such as AI company revenue or model costs without proof. No study is presented.
Interests and limits
Commenters are self-selected, and many are software developers, so the debate drifts toward programming. Claims about costs, model progress and jobs are asserted, not checked. The post under discussion is not the source here. Only the first part of the thread was read, so later replies are not covered.
Connections
These links are this wiki's own reading, not claims the sources make.
- Automation Without Understanding: Why the United States must preserve mathematical capacity in the age of AI. Commenters say the program exists to produce scientists, not results. Park makes a similar case for funding mathematicians as machines begin to prove things. Both ask what research really outputs. Is it a proof, or someone who understands it?