Glitchvid
@Glitchvid@lemmy.world
- Comment on AI’s hidden bias: Chatbots can influence opinions without trying, study finds 2 weeks ago:
And what about when the AI owning class introduce intended bias?
It’s one the scariest outcomes possible: if people forego their reasoning and critical faculties for chat-bots — if you aren’t even the one thinking your own thoughts, who is?
- Comment on Redox OS has adopted a Certificate of Origin policy and a strict no-LLM policy 2 weeks ago:
It’s almost necessary at this point. At least some form of AI scraper prevention.
I had to take my public repos down a couple days ago, individuals and belligerents using botnets make blocking scrapers via normal means (user-agent/CIDR block) ineffective. So things like CloudFlare or Anubis are becoming necessary.
- Comment on Uber is letting women avoid male drivers and riders in the US 2 weeks ago:
I think it’s a perspective thing.
Men are less likely to perceive themselves as potential SA victims: so the relative subjective “chance” of false accusationsagainst them vs being victims themselves impacts their priorities.
- Comment on Boy I was wrong about the Fediverse 2 weeks ago:
Yeah, I mean that’s true of any social space though, if you say something agreeable (definitionally) you’re going to get agreement. If you view upvoting as consensus building (i.e “I like this” / “I agree”) it’s just a more concise representation of a reply saying as much.
But that is scrutable.
What becomes a problem is content getting surfaced/buried on non-scrutable metrics (typically engagement) — ragebait isn’t anything new, online or in societies. But when algorithms target content that gets engagement, ragebait is naturally surfaced in higher proportions. Often time such platforms completely bury content or make it impossible to find something not explicitly surfaced (YouTube search for example is widely known to be terrible here, FB rabidly buries comments on posts).
WRT communities, there definitely are instances and communities with very different rules, values and expected behaviors. Federation allows communities to pick and choose what other communities they think they’ll get along with. This includes banning individual remote users if they don’t follow local rules, or defederating entirely if other instances have drastically different values.
The federation model as described does well by my metrics. I can pick an instance that shares my values, participate in communities (in the Lemmy technical sense) that share them as well — and largely avoid or choose not to engage with people from communities (in the instance sense) that I don’t share values with. This is extending “freedom of association” to online spaces in a way that large platforms largely cannot and willingly do not enable.
- Comment on Boy I was wrong about the Fediverse 2 weeks ago:
I would say scrutability in itself doesn’t automatically make an algorithm good. “Demote everything that doesn’t support Trump” is perfectly scrutable but leads to a skewed discussion.
This is mostly getting into normative vs descriptive philosophy. If it’s scrutable that a site/instance is demoting everything non-aligned with a worldview; then on the Fediverse it’s users’ choice to leave (and part of ‘community values’).
In fact I would say any content boosting algorithm at all leads to skew and what you call sycophancy. That includes upvotes/downvotes that affect what posts users see first. So I would get rid of all that stuff and just show purely chronologically.
To some degree, yes. New Reddit is particularly bad about this, it actively buries unpopular replies (but it goes further, and doesn’t just use upvotes) — Software like Lemmy is better, you can easily set Sort by New or sort by Top as the default. There’s also no ‘Karma’ system that propagates across the site.
Sycophancy is a human trait, so it’ll always emerge in social systems; but normatively, our systems should not cater to these negative traits (e.g. Twitter).
- Comment on Boy I was wrong about the Fediverse 2 weeks ago:
For algorithms, anything that isn’t a straightforward scrutable way of presenting user content is bad, IMO.
Algorithms that promote engagement, monetization, and sycophants are bad.As for community of communities, that’s how the Fediverse works — you have a home instance which communicates with other instances. An instance has (nominally) rules, and expected conduct, and is often centered around a particular interest (game dev, programming, cities or countries, etc) then these communities interact with each other.
Having home instances with shared values and a subset of the entire userbase allows for recognizing and connecting with other “local” users. The same way people would trust their immediate neighbors more than random people from the city over. It helps form webs of trust, and establish natural networks.
This is how human society has functioned up until very recently — it’s what the brain evolved to do.We can see the consequence of systems that don’t respect that fact, sites that try catering to everyone and put us in the same tent, it destroys social regulation, you cannot possibly hope to explain yourself to tens of thousands of angry people on the Internet, nor should people be exposed to such vitriol.
- Comment on Boy I was wrong about the Fediverse 2 weeks ago:
It’s not the point of the article, but I think it nonetheless speaks to the power that the community-of-communities model provides.
The algorithmic content surfacing models are what primarily rot online interaction. Having all-encompassing is another cause. Letting people join communities with shared values, and those communities collectively deciding who they interact with, is a fundamental model of human societies since prehistory.
- Comment on Leaked Email Suggests Ring Plans to Expand ‘Search Party’ Surveillance Beyond Dogs 5 weeks ago:
Same.
I was sure the emotional manipulation tactic would be extremely effective. Guess it was a little too blatant, even for the general public.
- Comment on When a Weather Forecaster Needs a Fruit to Prove It’s Him: Another Signal of Ambient Trust Collapse in Tech 5 weeks ago:
A system is what it does.
- Comment on Discord will require a face scan or ID for full access next month 1 month ago:
Because people want fancy animations, images, videos, stylized text, etc. And the easiest way to accomplish that is to just use a browser under the hood.
- Comment on Best retro-styled/Frutiger Aero/Y2K/Aqua clients? 1 month ago:
With respect to CSS, there’s already several projects that recreate the Win 7 and Vista look.
- Comment on Developer patches Wine to make Photoshop 2021 & 2025 run on Linux — Adobe Creative Cloud installers finally work thanks to HTML, JavaScript and XML fixes 2 months ago:
Not a requirement, but a preference.
Onlyoffice looks like it might be good, I’ll give it a try.
Can’t stand libreoffice, it feels much like Office 2007 which was the worst version I ever had to use — fixed with 2013 and 2016, but libre hasn’t caught up.
- Comment on Developer patches Wine to make Photoshop 2021 & 2025 run on Linux — Adobe Creative Cloud installers finally work thanks to HTML, JavaScript and XML fixes 2 months ago:
There’s really only two programs that make moving to Linux very problematic for me, that’s Photoshop, and Word.
At least with word I can ultimately just sequester that into a VM, or learn a different document program if push comes to shove (RIP all my workflows for citations and templates).
But PS is pretty much non-negotiable, it needs GPU acceleration of a native environment to run well, and there just aren’t any alternatives that can do what PS does — I need real channel support (painting on channels, copying between them per layer, actual alpha support instead of naive transparency) and more. As much as I hate Adobe, PS is one of those tools that I just know intuitively, all the texture or photo manipulation work feels entirely natural, and I just don’t think I’m going to find that ever again.
So, if Linux people can get it working through Wine, it’s a huge relief that I can finally leave the Microslop ecosystem.
- Comment on Google Revisits JPEG XL in Chromium After Earlier Removal 3 months ago:
At this point if you’re going to use WebP you may as well just use AVIF instead, better compression ratio and the support matrix isn’t that different between them.
- Comment on In wake of Windows 10 retirement, over 780,000 Windows users skip Win 11 for Linux, says Zorin OS developers — distro hits unprecedented 1 million downloads in five weeks 3 months ago:
I’ve got an Intel 6900K 8-core X99 system. Also not compatible with W11, but serving me well.
The issue is even if I wanted to upgrade, that market segment is effectively dead; X299 and X399 (AMD) were the last real HEDT platforms. The only thing now is workstation tier boards, which are about $1K and processors to match
- Comment on The story of a crazy Half-Life 2 bug, as told by former Valve dev Tom Forsyth (Mastodon thread) 3 months ago:
it doesn’t seem like too much of a leap to replace that with Alyx grabbing a random length of pipe or chunk of rebar or something from the multitudes of trashed urban environments she traverses throughout the game.
Could’ve used the socket wrench Alyx was shown with in HL2 promotional material. It was originally supposed to be her counterpart to the crowbar anyway.
- Comment on Do Not Put Your Site Behind Cloudflare if You Don't Need To - Rik's Weblog 4 months ago:
One of the biggest appeals of CloudFlare (aside from DDoS protection) is they don’t charge absurd^[unless you’re big enough they think they can shake you down] prices for bandwidth egress. AWS, et al have honestly predatory bandwidth pricing, they could be 1/5 and still profitable.
So given the choice of paying $30 in AWS egress that could balloon to hundreds, or a flat $20/mo, it’s not hard to see why so many people choose CloudFlare.
- Comment on Is Kagy web browser worth it? 4 months ago:
If you search some people already have. kagifeedback.org/d/6842-non-ai-unlimited-plan/50
And… it’s not a priority for them, maybe they’ll charge you $5 to disable AI features though 🙃
- Comment on Is Kagy web browser worth it? 4 months ago:
It will take a prompt, then run multiple web searches to get relevant info, recursively if needed, and then give a meaningful response with citations.
Do you have an example of this you can provide verbatim?
I’m just curious; I think the one application LLMs might actually be viable for is exactly this kind of connection finding in a large corpus, and since I’m doing lots of research, I might actually find personal utility.
- Comment on Is Kagy web browser worth it? 4 months ago:
The AI assistant is backed by several different models which IIRC just call out to those providers (Op*n API, etc) and rack up tokens in the billing system: image You might be right that the AI cost is included when below the plan price — to that I have to say, give me a fucking cheaper plan that doesn’t implicitly include the cost of AI.
- Comment on Is Kagy web browser worth it? 4 months ago:
As far as I can tell, you have to completely disable all keyboard shortcuts or else when you press
Aanywhere that isn’t the search box you get dumped immediately into their AI assistant prompted with whatever you already had in the search bar.It didn’t cost me more than a few pennies, but on principle the several times that happened made me angry.
Apparently some of the news views in search are also easy to dump you into AI land. There’s community CSS add on that hides all that stuff now, but I wish the company would let me just disable the AI traps.
- Comment on Is Kagy web browser worth it? 4 months ago:
I use Kagi because the truth is all other corporate alternatives at this point are unusable swill.
That said, I do not like the company and disagree with their choices in many aspects.
For one, while they don’t force you to use AI features, there isn’t a way to explicitly turn them off for your account, there always the opportunity to rack up token costs if you accidently hit one of the AI buttons.
They still don’t run their own index, instead complacent to just pay the other search providers. Additionally, if you’re trying to escape Google… Kagi runs on Google Cloud Services.
There’s more complaints, and I’m sure others will chime in, but that’s my take.
- Comment on US Government Urges Total Ban of Our Most Popular Wi-Fi Router 4 months ago:
It’s funny because I’ve had this one particular issue with two Asus routers that I manage for family…
They use this plunger power button design, you push the button in and it toggle locks in to place, the problem is that after a few years whatever mechanism retains the plunger fails and it always springs back and keep the device from staying on. So far the solution has been to cram a paper clip down the housing to hold it in. I just find comedy in having to apply that fix twice.
- Comment on Presearch launches new NSFW Spicy Mode 4 months ago:
What gets me is they’re still planning on using the advertising monetization model.
- Comment on The AI that we'll have after AI (Doctorow) 5 months ago:
The sad thing is those GPUs are in specialized boards with specialized servers and cooling, and not really good at the kind of work consumers would want it for, assuming you could even get drivers. So all of those GPUs will realistically get scrapped if the bubble pops.
Maybe we’ll see the EPYC CPUs get sold secondhand at least, those are socketed.
- Comment on Google is blocking AI searches for Trump and dementia 5 months ago:
Kagi uses Bing as it’s primary search index AFAIK, so no Google search API there.
That said, it does run entirely on Google Cloud Services, which I personally find ironic.
- Comment on AI Coding Is Massively Overhyped, Report Finds 5 months ago:
I think we’ve tapped most of the mileage we can get from the current science, the AI bros conveniently forget there have been multiple AI winters, I suspect we’ll see at least one more before “AGI” (if we ever get there).
- Comment on Artificial Intelligence Spots Hidden Signs of Depression in Students’ Facial Expressions 6 months ago:
If you feel you are not properly sedated, call 348-844 immediately. Failure to do so may result in prosecution for criminal drug evasion.
- Comment on Why the video of Charlie Kirk being shot was kept on social media platforms 6 months ago:
I agree with this viscerally.
A lot of people are expressing sympathy for the people in the Kirk crowd, but honestly I think it might be a good thing for them — to see first hand what Kirk and the Republican rhetoric is actually advocating for. Maybe it’ll snap them out of the fantasies they have of culling “undesirables”.
- Comment on Bye Intel, hi AMD! I’m done after 2 dead Intels 6 months ago:
It’s fine, modern CPUs boost until they either hit amperage, voltage, or thermal constraints, assuming the motherboard isn’t behaving badly then the upper limits for all of those are safe to be at perpetually.
