I'm a computer science graduate with experience in embedded systems seeking to get a career in software engineering. In my free time I love learning all kinds of things, working on complex projects in various programming languages such as C++ and Rust, etc.
IMO any contract, waver, etc., shouldn't be legally enforceable unless the signatory has actually read it. It's always seemed to me to be one hell of a pathway of abuse (in a way) to just be able to bind someone to be legally required to do anything you want, for example, by just relying on them not reading the thing they signed.
And if you don't think the Republican's will use this distrust to their advantage, well, I have a bridge to sell you. It's a very common tactic by them: create a problem that never existed, then wine about said problem.
Really? I didn't know ISP-provided modems could even do that, I assumed they were like routers and isolated to the local network and what the ISP gives them access to. I just wish they would give me more than "well it's for security and it's corporate policy" or something, it just seems like some excuse, and their remedy is "buy your own router" (which isn't an option for me, too expensive).
Because a crime was committed, and a pretty serious one at that?
If I blow up your house or steel $100000 from you and we both resolve our differences out of band, should I just be allowed to go about my day like I never did anything, or should I be punished for the crimes I committed? If I am not punished, it makes a mockery of the law that is (supposed) to have protected you, and if it happens repeatedly people will start wondering why the law even should exist if it clearly and obviously doesn't work. Granted, this has yet to happen again, but if OAI isn't punished it sets a very bad baseline precedent: that if I just hack you with an AI model, it's a-okay, and you can't do anything about it because eh, it's all good man!
My ISP won't give me the router credentials because they claim that that would somehow risk their network interconnectivity or something. I have no idea how this makes any sense since a router (shouldn't) give me that level of access, but maybe I'm missing something?
This comment is full of both a lot of hyperbole and theoretical outcomes that Idk even where to begin. There for one is no guarantee that Anthropic will remain around long enough for them to get to Opus 9 let alone a better model for one. The same could be said for OAI.
And again, what is the problem with a platform wanting to not be flooded with a bunch of low-quality code that gets abandoned after it's done what the creator wanted it to do? I keep seeing this claim of these models being better than "most" developers but have yet to see any actual statistics cited to back up this claim which makes me think it's all just vibes.
1. You are not entitled to a website hosting your content. A website refusing to host you is not censorship in any form whatsoever.
2. I'm pretty positive they did this because of infrastructure reasons as a secondary reason to their mission. LLMs mean people can just pump out projects at light speed, and that costs money to both host and do things with on CI/CD. If you want that potentially to be lifted in the future, pay them so they are able to have that infrastructure.
Pretty much. Which is why I'm far more concerned about that. So many of the AI doomers are terrified of some super-intelligent autonomous AI doing something on it's own that wipes us out (and, mind you, they have no evidence that any super-intelligent AI would actually do that, other than si-fi, but that's a different topic). But there's nothing stopping us from doing something stupid like developing some super-intelligent AI that has yet to not have any reason to disobey the "help the user" and is subsequently socially engineered to do something horrible that ends up destroying a nation for example.
> That sounds more like a phishing attempt than anything a real company should send.
And yet companies do it all the time. Which is hilarious because they also will happily tell you to beware of phishing and scams, but they do the exact same things a phisher/scammer would do
Honestly I'm wayyy more concerned with social engineering attacks than some theoretically superintelligent AI. Social engineering is, IMO, the far worse of the too
I'm a bit confused about this. As in: how is what cut and paste does right now exactly problematic? And: how does this perform with assistive technologies? Text that is faded out or even invisible visually more often than not is still detectable by assistive technology because it is in the accessibility tree. (This also is what makes certain kinds of websites which try to look "fancy and clean" but in actuality have 6-10 panels all on one page which appear and disappear as you do things extremely hard to use by people with screen readers: the screen reader sees all of the different UI elements, even if completely invisible.) Does this "alteration" remove the "cut" text from the accessibility tree or does it remain? If so, this "fading out" of the region will appear to do absolutely nothing and will confuse disabled users if this feature were to get remade like this.
Fully agree here. I love MUDs. The shear level of creativity that you can have in an MUD is practically infinite, and MUDs are just so fun to play in general. Especially when you start adding in triggers and sound packs and all that
Maybe? I'm not too familiar with Pandoc's internals so I couldn't say. (I do know it supports templates so maybe you could pull it off like that?) Although I'm wondering if you could just have the JSON as input and then you use pandoc as the transformer and it generates the full HTML file? An interesting idea for sure.
This is so cool! When I initially saw the title I was wondering how this is different from SlideJS or similar, but this is really cool and now I'm wondering if this could (at least in theory) be integrated into something like Pandoc as an output option?
I also thought this was hilarious. The very "safety" measures these models have prevented them from doing... Something that is designed to increase safety?
This is nonsense. By this logic some random corporation should have total control over your computer and the inputs you feed it and the outputs it produces to ensure nobody who isn't "verified crazy" uses it. That's essentially what your saying.
These models are, ultimately, tools. I would never trust some random corporation (particularly one with a profit motive and hypocritical stance, which includes both OpenAI and Anthropic, to be clear) to decide what isn't and is considered "crazy" and who and who isn't "verified" not to be "crazy". Especially when these companies have time and time again demonstrated (1) that they cry wolf way too much which leads to nobody taking their claims about how "dangerous" their models are seriously and (2) incidents like this where OpenAI makes a claim ("Look at how dangerous our models are!") and then doesn't be smart and just... Slow the fuck down (and when testing these things, actually sandbox them properly, which obviously wasn't done here or this attack wouldn't have been even possible).
I had to read this twice because at first I thought it said that you replaced a $120 bowling system with $1600 in ESP32s and I was like "So you replaced a good-cost system with something more expensive why again?" :D
I suffered this same problem with Anthropic, ironically. Mainly with reverse engineering things. That would usually end up requiring a ton of context it had to remember, and then compaction would kick in and it would just forget half of it (that half being really important too), and even switching to Claude Code and having it document everything wouldn't really fix the problem. It was honestly quite sad because I've usually gotten better results with it, but it falls a part the moment you try doing something like that that's really complicated and requires a ton of different attempts to get it right.
> As for AI-assisted engineering going well, I think the jury is still out.
Anecdotally, AI-assisted engineering has helped me flesh out ideas or to learn extremely complicated APIs faster than trying to understand the docs (which usually are labyrinthine). MS COM ones, for example. I can go read the docs but it's easier to get a quick idea of what I need to do if I ask Claude to provide me an example of doing something specific with it, because MS's code samples (particularly their full ones in, say, the windows Desktop SDK repo) have always been annoying for me to wade through because I have to filter out a bunch of noise. I can't (and won't) try to guestimate "productivity" improvements though, but as an assistant AI has (somewhat) helped. I still do all the engineering work though. Along with it giving me tips on using more modern language features for languages like C++.
Because the youtube video essentially says what the studies do (and it's a developmental psychologist who's giving that talk too). I felt it would be a bit easier to digest than more than 6 separate studies that you'd have to read. But if you really want them here are at least 6 of them all essentially saying the same thing: https://www.techdirt.com/2023/12/18/yet-another-massive-stud...
Because the point applies to the subject of this thread too, that's why. People think that all of this is super easy to just ban by fiat (law) but the problem is that the "simple" solutions oftentimes have harmful downstream effects that must be considered. For example, let's hypothetically say that we ban "addictive algorithms". What even counts as an "addictive algorithm"? Does my Mastodon home timeline count as an "addictive algorithm"? How about the Bluesky feed? Does the HN home page count as one? (Someone could (reasonably!) make this argument too, given that HN does dynamically update it's home page depending on rankings of submissions to end-users, even if such an update requires a page refresh.) So what I was trying to say (and the reason I brought up age verification/estimation) is that if we take the simple path we're going to catch a bunch of entities up in the ban that we... Probably shouldn't be banning. But you are also going to suffer a lot of pain trying to (not) cause this problem too.
I'm pretty sure the point being made is that things like this are very difficult to actually encode into law in such a way that it doesn't cause significant damage downstream. Like, take the age verification/estimation laws that are going around. We see the big companies actively encouraging it because the laws will (ironically) benefit big tech rather than harming them in any substantial manner: companies like Meta have the money and resources to implement them. But everyone who isn't a super big powerful company will be harmed because they will be unable to comply (or will find it incredibly costly to do so).
> What should be changed is the exemption algorithmically curated feeds still get under section 230.
Uh... Have you actually ever read section 230? There is no "exemption" that algorithms get. Like literally none. The section is probably one of the simplest forms of legalese in the entire US code. Please read it before making claims about it.