• 38 Posts
  • 535 Comments
Joined 2 years ago
cake
Cake day: June 7th, 2024

help-circle









  • I’m sitting here reading these comments as the low-end Dell laptop I just picked up for software testing is booting up and updating Windows. For logistic reasons, had to pick one up today, so had the pleasure of dealing with Best Buy sales staff 🙄

    From powering it up, it’s been 1.5 hours with updates and multiple restarts. Half of it was spent showing a progress indicator with a carousel slideshow of all the great AI tools I have no interest in using. Then it insisted on signing in with a Microsoft cloud account.

    It’s been eons since I actually ran a fresh copy of Windows. Amazed people still put up with all this nonsense.



  • If you assume Level 5 autonomy exists, you don’t really need ANY control surface inside the car. Those just become useless artifacts. Steering wheels, gear shifters, gas/brake pedals, turn-signals, or rear-view mirrors.

    Really, even map apps. Those just convert to glorified ToDo apps, where you say where you need to be at what time and the AV figures out how to get you there. Showing your position on a map is just more of a curiosity, like the airplane in-flight scenes, giving you a rough sense of place and calmly reassuring you that you are on your way (instead of, say, plunging into the sea).

    You don’t really need car windows either. Could be replaced with screens that show idealized imagescapes to enhance the transportation experience, or a movie.

    Not requiring brake pedals is just the first logical step, if you want to let a machine drive you.

    For those who actually enjoy the driving experience, AVs are stupid wastes of effort and tech.


  • May want to consider setting up a private, on-prem system. That way, you can reliably enforce privacy/GDPR rules. You can also tweak the system to support local training, RAG, MCPs, etc.

    This way, the costs can also be controlled. It’s some capital investment in local hardware, plus reasonably fixed power/cooling/maintenaance ongoing expenses.

    Another way is to use the major AI services for planning/brainstorming specific features, tell it not to implement or touch anything, but to generate a detailed plan for an implementer LLM. Review that plan manually, and when ready, feed it to your local system for implementation and debugging.

    This doesn’t work if the goal is one-shot vibe-coding. But it works really well for focused feature enhancements, test coverage, and bugfixing.