luciole

luciole , 4 days ago

Just please don’t be smart ass and choose a non relevant top-level domain because it looks cute. .io is for the British Indian Ocean Territory. .af is for Afghanistan. queer.af actually got taken down by the talibans.

.com and .org are both open TLDs and totally fine. If you’re afraid to be understood as organization, you can go for .com. It’s the default of the web by now.

If your service can be understood as some kind of web application, you could look into .app as well.

luciole , 4 days ago

The .io TLD has been the subject of controversy for a number of years despite (or because it is?) being hijacked by tech.

EDIT: More about it

luciole , 4 days ago

No, .com is not meant as commercial anymore and it was always open to everybody. No matter how easy the domain resellers are making it, picking TLDs has some implications: https://en.m.wikipedia.org/wiki/List_of_Internet_top-level_domains

luciole , 12 days ago

Signs get more severe as OP heads back home. Housing crisis and so on.

luciole , 12 days ago

Nah it’s biohazard. If I had to guess I’d go for upcoming zombie apocalypse ground zero.

luciole , 12 days ago

They sure like to use nondescript generative AI pictures. Can’t figure out what’s under the hood of their "Code Teacher" LLM. Most lessons are behind login.

https://beehaw.org/pictrs/image/c1894d81-0ee9-4c87-a5ae-1ef2d90d42d7.webp

luciole , 18 days ago

I want to ditch Windows, I really do, but when I get free time I want to either play a game or tinker on some side project. I don’t want to fiddle with drivers and what not for my OS. A year ago I killed a few weekends trying to get a Ubuntu partition nice and cozy for gaming but I got fed up fighting with all kinds of issues on basic things. The fact that games actually running correctly on Linux is hit-or-miss as well... It’s a hard sell (even though it’s free). Microsoft seems to be hell-bent on convincing me to try out some other Linux distro at some point though.

luciole , 18 days ago

I do! You mean like an actual second physical drive? Does that bring advantages compared to partitioning a single drive besides the space?

luciole , 19 days ago

Reducing emotion to voice intonation and facial expression is trivializing what it means to feel. This kind of approach dates from the 70s (promoted namely by Paul Elkman) and has been widely criticized from the get-go. It’s telling of the serious lack of emotional intelligence of the makers of such models. This field keeps redefining words pointing to deep concepts with their superficial facsimiles. If "emotion" is reduced to a smirk and "learning" to a calibrated variable, then of course OpenAI will be able to claim grand things based on that amputated view of the human experience.

luciole , 20 days ago

Was asked some months ago, unfortunately not much came up.

luciole , 24 days ago

Wrong article?

luciole , 1 month ago

The actual research page is so awkward. The TLDR at the top goes:

single portrait photo + speech audio = hyper-realistic talking face video

Then a little lower comes the big red warning:

We are exploring visual affective skill generation for virtual, interactive characters, NOT impersonating any person in the real world.

No siree! Big "not what it looks like" vibes.

luciole , 1 month ago

From the org’s definition of bots, I’d say it’s implicit that bot activity excludes expected communication in an infrastructure, client-server or otherwise. A bot is historically understood as an unexpected, nosy guest poking around a system. A good one might be indexing a website for a search engine. A bad one might be scraping email addresses for spammers.

In any case, none of the examples you give can be reasonably categorized as bots and the full report gives no indication of doing so.

luciole , 1 month ago

Yeah, their reporting suffers from not adequately defining what is being measured.

luciole , 1 month ago

Can you start by providing a little background and context for the study? Many people might expect that LLMs would treat a person’s name as a neutral data point, but that isn’t the case at all, according to your research?

Ideally when someone submits a query to a language model, what they would want to see, even if they add a person’s name to the query, is a response that is not sensitive to the name. But at the end of the day, these models just create the most likely next token– or the most likely next word–based on how they were trained.

LLMs are being sold by tech gurus as lesser general AIs and this post speaks at least as much about LLMs' shortcomings as it does about our lack of understanding of what is actually being sold to us.

luciole , 1 month ago

Does anyone actually use offline installers on a regular basis? I tried a few times and I had problems. Dunno if just bad luck. Never managed to install Pillars on eternity with it because it errored out every time. Another game's offline installer (can’t remember which) would stall for hours then crash. I suspect a lot of users would be in for a surprise if they actually tried them.

luciole , 1 month ago

Good to hear, I’ll check it out again and make sure I’m not having an issue on my end.

luciole , 2 months ago

JJ: So when you go in at a state legislative level, what do you concretely ask for? What is that language in the bills that you put forward?

GG: It’s actually pretty much consistent. There’s really only one active sentence, and it says that, “Hey, Mr. Manufacturer, if you want to do business in our state, you must provide all the same materials for purposes of repair that you’ve already created for your own repair services.” That’s pretty much it.

I feel like this is not enough. The counter would be to make the product not repairable at all and offer no repair services whatsoever, which is sometimes the case already.

luciole , 2 months ago

Apple sells an high end experience, not repairability. They don’t care about generating e-waste. AirPods for example are notoriously unrepairable. It’s just a matter of profitability. They’ve put in place a parts pairing technology to maintain a stranglehold on parts availability so as to circumvent legislation on repair. The obsolescence is planned. From the article:

JJ: So what does that malicious compliance look like? It’s a rhetorical support for the right to repair, but when it actually pans out, it doesn’t look like what you’re actually calling for.

GG: Yeah, the best example right now is what we call “parts pairing.” That’s been a problem all along, and we thought we had it nailed down in our template legislation, which we wrote back in 2015, that you can’t require specifically that you buy a part only from the manufacturer, and only new. And Apple got around it. They just said, “Well, we’re going to make sure that if you order a part from us, it’ll only work if you give us the serial number of your phone, and we preload that serial number into the part that we ship you, and that’ll work, but nothing else will.”

luciole , 3 months ago

It's kind of bleak. The web was supposed to be for everybody. I hadn't realized that in the last two decades we had lost the ability for neophytes to chug out HTML pages from desktop in a visual manner and upload them to a server for the world to see. Only non dead software I found that came close was Pinegrow, but it's proprietary.

luciole , 3 months ago

Apparently they have!

luciole , 3 months ago

I know you’re not alone with the opinion that a website asking an email address to create an account is dangerous, but frankly I still don’t understand the slippery slope argument attached to it. There are laws governing email marketing nowadays (CAN-SPAM in the US), as any actual business fucking around will find out.

In my humble opinion, an important lesson we can take from the last decades of the web is to be wary of a private free lunch. The Google search engine has never required an email and yet today they sit on an empire based on the exploitation of our data. In that sense, paying for a service is much more honest than mining the users’ privacy and selling it to advertisers (as mentioned in some hermetic Terms of Agreements & Conditions). The system may not be perfect, but asking an email address is the least invasive way to recognize someone that paid for a service.

Also, what do you mean by "self-protectionism"? It sounds like a derogatory euphemism for "making a living". It’s fine for four journalists to live from their profession. I think paying human sized businesses for services is quite different than doing the same with disruptive, market devouring corporations.

luciole , 3 months ago

Man I hope Servo pans out. I'm hopeful since it's seen increased activity lately.

luciole , 3 months ago

Someone knows any good Firefox hard forks?

luciole , 3 months ago

I wish I shared your confidence. Mozilla jumped on the VR hype, then the Metaverse hype and now they're specifically betting on generative AI. It's leaving me feeling as suspicious as the article's author about Mozilla's latest ventures.

luciole , 3 months ago

It's a free account, like the one you made so you can write your comment. I'd hardly call it paywalled.

luciole , 3 months ago

For anyone wondering about how they'll eventually address financial sustainability if Stract takes off:

Stract is currently not monetized in any way, but its website says it will eventually have contextual ads tied to specific search terms but that it will not track its users, which is similar to the system DuckDuckGo uses. Stract also plans on offering ad-free searches to paying subscribers.

I'd pay for independent, non meta, ad-free search. I bet a more straightforward approach is more energy efficient as well. In the meanwhile the big tech are running a gazillion processes on our data to suck every bit of wealth they can out of our existence through their free (in it's littlest sense) products.

luciole , 3 months ago

Kagi is a meta search engine though. They just do calls to Google, Yandex, Brave, etc. cut the ad rot and sprinkle some secret spice on top.

EDIT: source, https://help.kagi.com/kagi/search-details/search-sources.html

luciole , 3 months ago

Personally I find the "there is no such thing as a real picture" argument facetious and dangerous. Filters, optimizing zoom and autofocus is not the same as convincingly taking someone out of a scene they were in or putting them in a scene they never were in. One is a purely aesthetic adjustment while they other purports false information. Samsung Generative Edit further trivializes the latter and leaves no indication of the manipulation.

luciole , 4 months ago

I had a colleague perform a similar experiment on ChatGPT 3. He's ecoanxious and was noticing how the model was getting gloomier and gloomier in accordance with him, so he tried something. Basically he asked something like "Why is (overpopulated specie) going instinct in (location)?" The model went on to list existential threats to a specie that is everything but endangered. Basically it naively gobbled the loaded question.

luciole , 4 months ago

"You wouldn't download a car?"

luciole , 4 months ago

They're going to make us miss that clueless little bastard.

luciole , 4 months ago

Robotics and AI algorithms make total sense in factories. I'm very surprised they're making them humanoid though. It's stated that'll make them dextrous or something, but I don't see it.