• 0 Posts
  • 86 Comments
Joined 3 years ago
cake
Cake day: July 24th, 2023

help-circle












  • AI fits very well into the “Workism” narrative. As LLMs generate the statistical medium of all the training data it has, the output is mediocre by design. That makes it an excellent tool for bullshit work, like creating an email summary of a meeting that no one will ever read, and countless other mindless and useless daily chores of the “knowledge worker”.

    But the basic premise hasn’t moved an inch with AI. If you have a job which rewards creativity and needs you to develop new ideas, solve hard problems etc. then AI doesn’t do shit for you. AI gives mediocre answers to mediocre people. It’s not creative at all. It spits out the average of the sum of its training data points. If your job consists mostly of doing mindless chores then it is not a good job anyway. Nothing to be proud of, not a career, just a job.




  • There is another problem which keeps stocks like Tesla afloat when it really should have tanked a long time ago: Index funds based on indexes like S&P 500 etc., where investors are automatically buying Tesla stock when buying into the fund, which stabilizes the stock in question. Musk knows this and that was the exact reason why he tried to get indexes to bend the rules of inclusion for spaceX to be included in indexes earlier than usual. The grave consequences are people investing in index funds are directly milked. Pension funds often include indexes, so what Musk is doing there is in effect trying to steal from people who did not actively invest in SpaceX.



  • These models generally are not capable enough to do one-shot vibe coding. They are pretty good as coding assistants if you tell them exactly what you want and let them focus on a specific aspect/part of code, not the whole thing at once.

    Using an agent framework (I like Kilo on VSCode but there are many others) you can start with a planning session to let the model find out what you want to build. Then you let it write that gist into AGENTS.md and double check if that is what you want. AGENTS.md will be loaded into the context automatically so the model has a solid base of understanding for everything you do afterwards. Once you have that, building in vertical slices on top of that skeleton is much easier. Another neat trick is to ask the model a few questions about the current code base (if there is any…) at the beginning of a session, e.g. "How does feature x work in the current code? ". This primes the model for what you are about to do. All this is obviously a bit more work than just vibe coding away, but it lets you keep in control of the code and helps in being alert for errors these models (and all LLMs in general) will inevitably produce.