• 1 Post
  • 64 Comments
Joined 3 years ago
cake
Cake day: June 13th, 2023

help-circle




  • Averages can mess this with this kind of statistics, where small group of people can bring up the average cost significantly.

    Gas is also expensive, my wife spends about $130 a month a gas alone. They are also likely factoring in all costs too, including personal property taxes (which where I live gets much more costly if the vehicle is worth over 20k), and all maintenance. So things like tire changes, replacement batteries, oil changes, and everything else, averaged over the lifetime of the car.

    You won’t see most of these with an electric car at nearly the same cost. Electric cars see much lower operating costs, but only if you can afford it, and can charge it cheaply. Many people I know can’t as they live in apartments and would have no way to charge an electric car.

    For us personally, looking at buying a new electric over my wife’s paid off car, increases in personal property taxes and insurance negate much if the financial savings of having a lower operating expenses. Combined with a high initial cost of the vehicles it doesn’t save anything financially.













  • Don’t forget the fundamental scaling properties of llms, that openai even used as the basis for strategy to make chat gpt 3.5.

    But basically llm performance is logarithmic. It’s easier to get rapid improvements early on. But at later points like we are now require exponentially more compute, training data, and model sizes to get now small level of improvements.

    Even if we get a 10x in compute, model size, and training data (which is fundamentally finite), the improvements aren’t going to be groundbreaking or solve any of the inherent limitations of the technology.