It’s mostly from people using their personal accounts to run LLM services that serve a larger team or organization. At least at Microsoft, it’s still impressively hard to get access to an LLM for service usage with high enough rate limits to be useful, making running services on dev boxes much more appealing (despite the countless drawbacks that few people seem to care about around security, compliance, reliability, etc).
It’s interesting that models seem, to a distant outsider of biology and drug discovery like me, to be good at coming to new conclusions from existing data. I feel like in coding, it’s the opposite - I have to drag the models kicking and screaming towards anything resembling a novel or nuanced approach to some problems. If anything, this behavior in coding is why we say senior+ engineers will continue to be high value employees, because we can steer the models away from boilerplate and overly generic solutions towards ones that fit aspects of the domain we understand more intrinsically.
This could easily just be how it looks from the outside of biology, but it does seem to produce more novel conclusions in biology than it does in coding and art. Curious if others have counter examples…
The change with this justification absolutely exists and has existed for a long time, but this justification only works for small changes (in the past and still today). Don’t come to me with a huge refactor or new feature and this explanation, which is happening more and more recently (it’s an easy strawman to make and hard to prove, but I certainly see it in practice).
Some of us don’t use an adblocker because we are aware that ads are how most websites get paid for on the internet. I pay for sites I care to support, but for something like LinkedIn, I’d rather have their monetization scheme push me away from any usage than block the ads and make the site slightly more palatable.
I think this is a pretty unfair takeaway when AI generated text is often so wordy, flowery, and full of jargon and abstract nouns. People are never going to read East of Eden or academic dissertations as quickly as direct, simple language either
You’re not a novice, there are a lot of us who know exactly what we are doing and see this as a huge downside. We are just being told to go faster, faster, faster lest we miss out on… something?
Both ahead of time compilers and JIT compilers often perform autovectorization of tight loops. The problem is that lots of hot loops are not necessarily simple loops, and in particular a lot of source code is written in a way which uses sequential dependencies that can’t be modeled in SIMD code. Aside from undefined behavior in C/C++, most compilers will fail to autovectorize because doing so would very slightly change the behavior of your code in a very hard to understand way.
I feel like this can be explained in part by “I like using it for myself, but I hate when others use it.”
When you use ChatGPT for yourself, you may have a sense that what you see is made up; when someone else that you trust uses it and pronounces the output in a way that suggests it is their own, you are left doing much more complex social math to figure out if your trust in this person or entity can hold. It gets exhausting, personally.
When AI produces those meaningful advances in those fields, great, we can start having meaningful discussions about them. The greatest medical advancement of the 21st century is likely mRNA, or maybe GLP-1 for some. Neither were LLM assisted in any meaningful way as far as I know (they predate ChatGPT, perhaps more primitive models were involved in ways I’m not familiar with). Until those advances come, this argument is fanfic.
Plus, in the most morbid way possible: who gives a shit about living longer if they are stripped of their career, are inundated with slop at every angle, and can’t trust any information. These are real problems that AI has already created, unlike the fanfic of ridding cancer.
I have no advice to offer, I only wish you good luck. I am still lucky enough to be employed, but when this whole parade ends, I have no idea what comes next - my only skill is programming and related knowledge work. I think the only path forward is to try to jump ship to another white or blue collar industry…
Have you considered just… writing code? Like we used to in the good old days? If the tool drives you to that point of frustration, maybe it’s time to give the tool a break.
> when the author is an AI and the consumer is a compiler, and no human needs to read the output at all.
This seems like a big bet on the assumption that fully autonomous codegen without humans in the loop is imminent if not already present - frankly, I hope you are wrong.
Even if that comes to pass in some cases, I also find it hard to believe that an LLM will ever be able to generate code in any new language at the same level with which it can generate stack overflow-shaped JavaScript and python, because it’ll never have as robust of a training set for new languages.
That seems correct, but it also doesn’t account for managed languages with runtimes like JavaScript or Java or .NET, which probably have a lot of interesting runtime info they could use to influence caching behavior. There’s an amount of “who caches the cacher” if you go down this path (who manages cache lines for the V8 native code that is in turn managing cache lines for jitted JavaScript code), but it still seems like there is opportunity there?
Sure, but this is based on a fundamental trust in governments ability to spend money effectively. The ineffective spending has been in the news way more than the effective spending, so some people take this to mean all of the spending is ineffective.
I don’t know how to square this skepticism of government against very vocal “patriotism” coming from the trump camp, but humans can contain multitudes, I guess?
In C#, you would normally implement rules like this with a custom Roslyn Analyzer or with https://github.com/dotnet/roslyn/blob/main/src/RoslynAnalyze.... It’s fair to say C# projects tend to have smaller denylists than mature C++ projects, but banned APIs definitely exist in mature C# projects.
I didn’t know this, but there are also security downsides to being ahead of chrome — namely, all chrome releases take dependencies on “known good” v8 release versions which have at least passed normal tests and minimal fuzzing, but also v8 releases go through much more public review and fuzzing by the time they reach chrome stable channel. I expect if you want to be as secure as possible, you’d want to stay aligned with “whatever v8 is in chrome stable.”
If people aren't getting their work done, then they should be having discussions with their manager that eventually lead to pip or firing if not resolved. If they are getting their work done... Who cares if I do a "non work thing" at a "work time"?
And also, a decent chunk of alcohol consumption must be solo? I'd bet alcohol is broadly more social, but I would also wonder if that would change if more public gathering places served weed in some form.
There is a ton of tools and custom logic used by/with/for the GN ecosystem in chromium that I imagine would be difficult to port.
This tool is substantially less complex than Bazel, nor is it a reimplementation of Bazel. Ninja's whole goal in life is to be a very fast local executor of the command DAG described by a ninja file, and siso's only goal is to be a remote executor of that DAG.
This is overall less complex than their first stabs at remote execution, which involved standing up a proxy server locally and wrapping all ninja commands in a "run a command locally which forwards it to the proxy server which forwards it to the remote backend" script.
A handful of other areas are configured using Starlark in chromium. This particular use is in a very different capacity than Bazel - the Bazel equivalent in chromium is GN, and I have not seen any signs that GN will be replaced any time soon.
Saying that Microsoft is "Rewriting Windows in rust" suggests you might not be as informed as you think... Very specific components with history of performance or security issues are getting ported in a very uncoordinated effort. Windows will be primarily C, C++, and C# for a very long time to come
This isn't a github runner issue, it's an Azure VM issue. They use Azure VM SKUs with temp disks, and windows VMs on those SKUs by default spin up with C on a remote-backed, persistent file share and D on the local temp disk. The remote-backed file share OS disk is absolutely slow, especially if they're using Standard HDD or Standard SSD disks. You can spin up your own VMs that use Premium OS disks for slightly better performance, but if you are doing anything serious on an azure VM, you should use the temp drive (which, by the way, mounts to /mnt on Linux) or ephemeral OS disks.
(And also this is all for v5 and earlier skus and changes slightly for v6 skus but whatever).
Surely there is a middle ground for contributions which you don't really care to retain ownership of? I don't sign CLAs for projects I want to form a long term contributor relationship with, but if I am just trying to fix a small bug that the (probably corporate) owners don't care to fix themselves, I'll sign that code away without hesitation.
Someone already said The Rest is History, but one of the presenters of that podcast Tom Holland (not the actor) has also written extensively about the history of the catholic church in Millennium and Dominion. Highly recommended.
Which take millions of dollars and huge teams to make. These take one bored person, a sentence, and a few minutes to go from idea to posting on social media. That difference is the entire concern.
Is there a way to get phones to you or information about how to do this myself? I am sure I could find similar programs if I looked but it sounds interesting.
I don't think it's true that C# isn't taken seriously at Microsoft - some of the frameworks on top of C# come and go, but I worked on JS on WinRT briefly and can say it was never, ever "taken seriously" in the way the core C# language and runtime is. If you want to work with azure or windows, C# will be the obvious best choice for the foreseeable future, and if you want to work with C# outside of those verticals I don't think it's a bad or dead end choice either.