sure, that's "what people want" inasmuch as if you put every button through a statistical microscope, that's what the statistics will tell you, but if you give a rat cocaine-dispensing button and measure how many times it hits the "more cocaine" button you'll also come away with the conclusion "rats want cocaine", a thing they never encounter in nature and would never have encountered without you putting it in front of them, and you'll pat yourself on the back and say "now I understand rats: they are all vicious cocaine fiends", but you haven't really learned about rats' true nature, you'll have only conned yourself into a false narrative that confirms that your own actions are only "giving the rat what it wanted", and after it dies of an overdose, you declare yourself innocent. Anyway that's a/b testing and the tech industry.
It's been some time since I checked in on student related stuff, but..
IIRC PSDs contain some process related information that instructors check on; like for example photoshop layering contributes to file sizes and they don't want their students abusing it to the point of large file sizes; it'd look bad for their school's reputation.
I think the point is that LLMs makes it easier and cheaper to produce a large volume of convincing lies. The candidate likely would not have been able to produce convincing-enough lies to get through the resume screen without LLMs.
That's true. On the other hand I have tried ChatGPT to review programming concepts or language features and I have found it very convenient and more useful than Googling.
For instance if you want to prepare for a C dev interview and would like to review what 'static' means and does (one of the super usual interview questions) you can just ask and immediately get a pretty much perfect explanation without noise. It's not cheating, it's just a better tool.
How do you reconcile that opinion with the fact that LLMs trained on programming concepts generally give incorrect answers about 50% of the time?
Is it actually more useful than Googling, or is it just so convenient that you let it convince you that it was useful? Or, depressingly, is Google just becoming so useless that something wrong a solid half of the time is still better?
I have a young colleague who wanted a job at a FAANG company, and asked for advice. I said spend a couple weeks studying the leetcode books - it will be the best value for time spent you'll ever get.
Or you can open any good C book and review that way. Not to bash on the use of AI, but there's a lot of alternative ways that for me is more reliable to get knowledge from.
I'm not sure that it's a good thing if "ability to produce convincing lies" is something that a company requires in a job candidate. People getting into jobs who aren't exceptional liars when they couldn't have otherwise seems like win to me.
on the contrary, they met after John went to one of Yoko's art shows and was intrigued by her artwork; it's probably unlikely that Yoko would have married John if it were not for Yoko's work.
people love to talk about a game being realistic or not realistic without asking whether more realism = more fun. Usually the answer is no. Doing collision detection on every bullet on every frame is a cost that may or may not be worth it, depending on the overall design of your game. "Games take shortcuts" is really flattening the discussion and acting like the game developers don't know these things that the very smart blogger knows, when in reality, game developers know this stuff and make conscious design decisions. Most games are not designed to be scientific modeling programs, they're entertainment products, governed by questions about entertainment value, not simulation accuracy. Plus in real life you're not going to receive a network packet that says "a bullet whose trajectory will hit you in the skull will reach you in 0.7 seconds" and give cheaters enough time to auto-duck with a script. Even if you can afford to do collision detection on every bullet on every frame, there are many other reasons why it might not be a good idea. Hitscan is a design option, not something that's "right" or "wrong".
critics say that, but its viewership is low compared to other dramas. Usually when I tell people that it's one of my favorite shows, their response is that they've never seen it.
^ compare the viewership. The Americans was getting fewer than a million viewers in its final seasons, Lost regularly got over ten million. You can't easily chalk it up to just the fact that Lost was on network; The Last of Us was getting more than 10x the viewership of The Americans and it was on HBO. While The Americans is very critically acclaimed, it is not as widely watched as it really should be.
There must be a disconnect between viewers and critics, because viewership declined over its entire run. Compare that with a true masterpiece like Breaking Bad which saw viewership consistently increase over the course of its run.
Pretty rare for any show to increase in viewership year over year. Breaking Bad had monumental word of mouth towards the end. Better call Saul dropped year over year but was still fantastic.
Go doesn't have a ? operator today, and the ? operator being used for error handling has precedence in Rust and Zig, so it doesn't seem to be all that out of the ordinary or without precedent in other languages.
The proposed behavior is the opposite of how Swift and Rust use ?. `foo?.bar()` invokes `foo.bar()` in the non-error case, while in Go `foo ? { bar() }` invokes it in the error case.
there's some precedent in the direction of adding predeclared identifiers for error handling: the identifier `error` was originally not predeclared, you had to import it from `io` (or maybe `os`?) and refer to it as `io.Error` everywhere.
> The idea of opening another bank or broker, despite how simple it is
I would wager that there are a lot of people that have money in Vanguard because Vanguard provides their employer's 401k, and their choice was "Vanguard or no 401k", not "Vanguard or something else of your choice".
Vanguard is a good choice for managing retirement funds. If you want to speculate on high-risk investments, whether doctoms or crypto or whatever, it's better to do it outside your 401k.
The argument that you are making is that a single entity controlling the pricing of a good is bad. That's not a counter-argument, it's the same argument. It's bad when a government does it, it's bad when an oligopoly does it. More than one thing can be bad.
reply