The two major US parties are in the midst of another swap of which is conservative and which is liberal.
Marin County California, probably the area most heavily voting for the Democrat party, is clearly the most classically conservative part of the country, allowing almost no development and strongly objecting to even the slightest offenses in speech, whereas rural counties in the south want classically liberal safety nets and protections and heavily vote for the Republican party.
He's writing in the academic/upper-class style that it's training data focused on.
I'm convinced that one of the largest frictions to common use of LLMs is that it translates everyone's writing to the that same style. Having it punch-up or flesh-out your proposal or outline, or whatever, isn't really adding any new material, but it's translating it to a writing style that has historically been exclusive and difficult to learn, and that difference in style is what made the original text sub-par, from the perspective of academia and members the upper class.
Copyright DRM tries to prevent unapproved recording, and has completely failed, with effectively every commercial release with any interest being available of peer-to-peer file sharing networks.
Videogame anti-cheat measures try to prevent assistance in real-time games, and have created a cat-and-mouse game of alternately succeeding and failing. For a non-real-time games, detection would be far more difficult.
The best-case scenario for preventing the release of questions and the detection of cheating in an online test is that the questions will be released and you might sometimes detect some cheating.
Pretty much every successful low-to-mid range phone in production has larger batteries, not smaller ones. Apple completely missed a market that is very easily researched.
The text on the buttons cuts off after the first word, mangling most of the options.
Every decision it presented to me, except Marbury v. Madison are presented as expanding or limiting rights, when in reality they may be preserving rights or weighing one group's rights vs anther's, so defining them as only expanding or limiting does not always make sense.
Inference (i.e. running an already trained LLM) requires current top-end hardware, but over time that will become commodity hardware, so there's only a small perioud (perhaps the next decade) where running an LLM as a service instead of locally even makes sense, so they will likely not go away because they are too expensive, but because they are too cheap.
Neural network development platforms are even more bloated and broken than the record set by FPGA development platforms and even mobile phone development platforms.
reply