HN user
Doing the things a Rand-man can
I think small children learning language learn in a way similar to LLM's, just saying words that seem appropriate to the situation without any understanding of the meaning. But a key difference in my mind, is they use the reactions to this text prediction to help them construct a picture of the world, which then can interact with their thinking of what words to say.
I feel like this might be the transition between an industry driven by apprentice-ship like guidance to an industry driven by credentialed training.
This reminds me of actors or entertainers who clearly are going through a manic phase that is veering towards psychosis or a crash.
Sometimes leave Britney alone means actually leave her alone.
This is also an option with AI patches, if your AI contribution is good enough, let it get proven in use, and then it can be considered naturally as the developers of the original project look at what's going on with similar projects.
Code that works but is unmaintainable is a burden on a project, that being said with open source, you can always fork the code, add your code, and let people decide
One factor here is how systems become more or less prone to creating great or mediocre men. Inbreeding, isolation, over romanticism of emotions, all these are factors that can make a dangerous inconsistent person more likely to appear in the halls of power
A little off topic but my position is that a Linux distribution ought to be a series of build options, package selection and configuration. Ultimately, you should be able to do this in a few files.
The fact that it takes so much more to put together a Linux distribution is I think something that the Linux ecosystem ought to aspire to change.
You can create very sophisticated simulations all you want, but in the end, you can only test the accuracy of your simulation when you try the real thing
Because it's international waters
I think eventually you are going to end up with every smart AI continually checked by dumber AI's to make sure they don't do anything too crazy. Which probably does bring AI closer to how human intelligence works
This is something that intrigues me: "Is able to output internal representations (tokens, abstract syntax tree, intermediate representation) in machine-readable JSON form."
I'm not sure what I can do with that information but I wish I had an excuse to use it
That's interesting, it reminds me of something we've realized internally at my company, AI coding is best used with strict adherence to requirements and tests (potentially generated by AI), reviewed by a human developer
I don’t mind using Wayland these days, but I do feel most of the security arguments are aimed at the government/corporate/big server audiences than the single user/developer.
Which ultimately is fine, this reflects the focus of the people who have the skills and opportunities to contribute and is unlikely to change any time soon.
That is somewhat unfortunate for some but ultimately if you’re asking people to work for free you can’t be too picky on what they choose to work on.
My family has typically stir fried them, chopped up with olive oil salt and pepper, so I always got confused by Brussels sprouts horror stories
I think a good manager should also be a cushion between the higher up politics and his team, so they do perhaps get more praise than is deserved for their team's successes but they should also absorb much of the criticism for their team's failures
This is a grey area, look at some countries where you’ve had protesters block aid trucks to refugee camps saying that their national is imminent danger from these people. I think a key here is the degree and length of traffic blockage. When it is being used to essentially lay siege to a place it is ambiguous whether it’s violent or non violent. Temporary limited disruption of traffic is probably pretty firmly in the non violent column
This isn’t the point of the piece, but I have found that the thinker often gets in the way of the builder, because there’s always a better way to build, there’s always some imperfect subsystem you just want to tear out and rewrite and then you realize you were all wrong about this and that, etc.
More to the piece itself, I know some crusty old embedded engineers who feel the same way about compilers as this guy does about AI, it doesn’t invalidate his point but it’s food for thought
The upvotes are interesting because the comments are either tangential or negative, so maybe ai’s look at upvotes as this is neat to look at but I don’t necessarily agree
I feel like the arms race between student cheaters and teacher testing has been going on for hundreds of years, ever since the first answer key written on the back of a hand
It would depend on your accessibility needs, if you only need the apis included in a, then a would be a better option
I think if he had died at 50 his reputation would be much much better, although he is certainly not unique in that regard
Say it ain’t so Joe
This is probably my approach (maybe minus the AI), I do think networking and reaching out to people can yield better results but I was never any good at that
This has bit me a few times when a Linux system crashes so there’s no final call to fsync implicit or otherwise
Main issue author had with lambdas is autogenerated names in crash reports
I’ve started with a cpap recently and my biggest issue is that it interferes with snuggling with my wife, that may seem trivial but it can have a big impact on my mood
I'm curious if anyone has a good solution for this, how to accurately display cases where a storm might hit either Monday or Tuesday, say 50% chance each. If you just say there's a 50% chance of rain on both days, it looks like there's a significant chance (I guess 25%) that it will rain both days, when the real likelihood of both days raining might be far lower.
I think when we look at general intelligence we need to look at multiple methods of reasoning working together. Looking at children there is an instinct to blindly imitate those around them, trying to match behavior without fully understanding it, however the experiences from this behavior then feed into a symbolic understanding in the child’s mind, allowing them to reason about the situation later on
Here’s the question, are the benefits worth the increase in complexity, or rather are they worth more than other features that could be worthwhile for libraries to support.
For the hyperscale web sure, but for the long tail web that is very unclear.