Sailor, software engineer, musician, terminally online.

Englishman in Wales.

  • 0 Posts
  • 24 Comments
Joined 3 years ago
cake
Cake day: July 5th, 2023

help-circle
  • It’s not just about how many examples of a language exist in the LLM’s distribution though, dynamic typing is a poor choice if you’re using LLMs for software in my view because it turns compile-time errors into runtime errors which will only be caught if the LLM writes a suitable test; an area that they still leave a fair bit to be desired in.

    The whole virtue of dynamic typing is that it reduces cognitive load on the programmer, if you’re using LLMs there’s no reason to do that in my opinion, you’re offloading a bunch of cognitive load to begin with so expressiveness and constraint is more important than human ergonomics. Yes there’s loads of JS in an LLM corpus but a lot of it is going to be poor examples, because it’s generally quite easy to write JS badly. If you use typescript you’re at least forcing a few common hallucinations to produce compile-time errors which the model has immediate feedback for, and you’re giving the model more context on what it’s seeing.






  • I have a neurological vision disorder which I suspect augmented reality glasses would help a lot, but sod letting Meta see my life from a first-person perspective 24/7, or the lives of anyone around me for that matter. Also I suspect a genuine medical device would have to be made very obvious, you don’t want accessibility tech being mistaken for pervert tech. I wouldn’t take the risk frankly! It’s like if I’m talking to someone and they have a hearing aid in that’s fine, but if they’ve got headphones in I’m going to feel a little cheesed off on the face of it.

    Also not American but I work for a trans-Atlantic company, those things sound like a PHI nightmare in a medical setting given Meta’s habit of indiscriminately hoovering up data like as many labradors in a pantry.













  • What worries me is people using it for safety-critical things without appreciating the danger, for a test some of my friends who sail for a living asked various chatbots some questions like ‘how would you approach a berth under sail when the wind is against the tide?’ and every single one but Claude got it dangerously wrong, and Claude wasn’t 100% either it just refrained from suggesting a sail plan that would lead to you crashing into the pontoon.

    Will give a shout out to Mistral which got the sailing element of the question very wrong but did sensibly suggest ‘turn your engine on and perform the manoeuvre under power’ which of course is what you should do if you can.



  • Why is stirring up conflict here beneficial though? I’m intensely critical of the current US government but me and the average American still have more in common with each-other than either of us have with our respective ruling elites in terms of material interest and daily life.

    Yes their crowing about being the most democratic country on Earth is a bit silly but that doesn’t mean they deserve to become an outright plutocracy, nobody deserves that. You can object to Washington’s lofty perch in the global pecking order and to Americentrism in general without wishing ill on a whole nationality the vast majority of whom have done nothing to wrong you personally. The only thing I wish for the Americans is for them to fix their backsliding democratic institutions and to adopt a more realistic view of their country’s place in the geopolitical landscape; I don’t doubt these are goals many Americans support at this point.