Jesus_666
@Jesus_666@lemmy.world
- Comment on If AI was going to advance exponentially I'd of expected it to take off by now. 3 days ago:
AI isn’t taking off because it took off in the 60s. Heck, they were even working on neural nets back then. Same as in the 90s when they actually got them to be useful in a production environment.
We got a deep learning craze in the 2010s and then bolted that onto neural nets to get the current wave of “transformers/diffusion models will solve all problems”. They’re really just today’s LISP machines; expected to take over everything but unlikely to actually succeed.
Notably, deep learning assumes that better results come from a bigger dataset but we already trained our existing models on the sum total of all of humanity’s writings. In fact, current training is hampered by the fact that a substantial amount of all new content is already AI-generated.
Despite how much the current approach is hyped by the tech companies, I can’t see it delivering further substantial improvements by just throwing more data (which doesn’t exist) or processing power at the problem.
We need a systemically different approach and while it seems like there’s all the money in the world to fund the necessary research, the same seemed true in the 50s, the 60s, the 80s, the 90s, the 10s… In the end, a new AI winter will come as people realize that the current approach won’t live up to their unrealistic expectations. Ten to fifteen years later some new approach will come out of underfunded basic research.
And it’s all just a little bit of history repeating.
- Comment on Don't Look Up 3 days ago:
Fair enough. The dumb ones were just loud enough to sound like 90%. But I sure did talk to some panicked dumbasses online back then.
- Comment on Don't Look Up 3 days ago:
I agree with the “nutcase” diagnosis but I think she just can’t tell paragliders and parachutes apart and thought these were paratroopers. Which Hamas somehow has now.
- Comment on Don't Look Up 3 days ago:
I remember the early 2000s when basically 90% of all Americans were absolutely certain that jihadists were going to attack their local supermarket any minute now because Power Cable, Nebraska was such a strategic target.
Heck, there was a bomb scare because of an advertisement campaign for Aqua Teen Hunger Force that involved placing PCBs with LEDs on them that would display characters from the show. Because surely Al Quaeda would put conspicuous LED displays on their bombs.
News media want people to panic so they keep tuning in. Panicked people tend to come up with remarkably stupid scenarios like “Al Quaeda have unlimited resources and can show up anywhere to shoot people at random” or “Hamas want to take Dorcester as a strategic location to strike at Israel from”.
- Comment on Realtek's $10 tiny 10GbE network adapter is coming to motherboards later this year 1 week ago:
Ever since the BE200 debacle I don’t know if I can trust Intel to deliver. Sure, the stuff that’s already out there works but who knows if any of their future stuff will?
- Comment on We did the math on AI’s energy footprint. Here’s the story you haven’t heard. 2 weeks ago:
AI usually got better when people realized it wasn’t going to do all it was hyped up for but was useful for a certain set of tasks.
Then it turned from world-changing hotness to super boring tech your washing machine uses to fine-tune its washing program.
- Comment on Github Discussion: Allow us to block Copilot-generated issues (and PRs) from our own repositories #159749 2 weeks ago:
Why not submit a million vibecoded PRs against various Microsoft repos? At some point they might get it.
- Comment on Bungie appears to have plagarized an artist's work and style in their extraction shooter "Marathon" according to this Bluesky post 2 weeks ago:
My vent core feels retroactively unblasted by this revelation. You really can’t expect anything good from any of the major studios anymore these days.
- Comment on xkcd #3089: Modern 2 weeks ago:
Wasn’t there a Dark Age after Bronze? The one where everyone was scowling the whole time and the stories were so tryhard edgy you could use a typical Youngblood issue as a letter opener?
- Comment on The clueless people are out there among us 3 weeks ago:
That’s quite how AC outlets work. Line and neutral can intentionally have different potentials relative to ground depending on how the house’s electrical system is designed. This can become relevant in certain situations like very simple devices (think “lamp socket with a power plug”).
A plug that can’t be inverted makes this a non-issue.
- Comment on The clueless people are out there among us 3 weeks ago:
I think the Swiss have the best Europlug-based system. Their three-conductor plugs have the same footprint as basic Europlugs, which makes for very dense plug arrangements. Unlike e.g. the German Schuko plug they only fit in one orientation so you get no polarity issues.
It’s pretty neat.
- Comment on Ditch the DIY Drama: Why Use Fedify Instead of Building ActivityPub from Scratch? 1 month ago:
Of course you wouldn’t use an existing database engine as the foundation of a new database engine. But you would use an existing database engine as the foundation of an ERP software, which is a vastly different use case even if the software does spend a lot of time dealing with data.
If I want to build an application I don’t want to reimplement everything. That’s what middleware is for. The use case of my application is most likely not to speak a certain protocol; the protocol is just the means to what I actually want to do. There’s no reason for me to roll my own implementation from scratch and keep up with current developments except if I’m unhappy with all current implementations of that protocol.
Of course one can overdo it with middleware (the JS world is rife with this) but implementing a communication protocol is one of the classic cases where it makes sense.
- Comment on Spain public broadcaster calls for 'debate' over Israel's Eurovision participation 1 month ago:
I’m sorry but that’s Germany’s job already.
- Comment on Always explains so much 1 month ago:
Remember that each woman has a different experience. Some women have such a light period that they only notice it because they find spots in their underwear. On the other hand, a woman with endometriosis will probably start being in excruciating pain the day before the period starts and will know exactly how long it lasts.
Source: I know someone with endo. She also has a resistance to at least one OTC painkiller because she used to pop those things like candy. And a female coworker of hers (with a very light period) thinks that most women are exaggerating and that periods aren’t much of a problem…
- Comment on We are so cooked 2 months ago:
Usually, when people talk about bees dying, they mean wild bees. Unlike honey bees they aren’t cultivated by us. They also tend to be better pollinators than honey bees, adapted to local plants that honey bees can’t handle well.
- Comment on With the current state of the news, April's fools aren't fun anymore because they can't be distinguished as easily as before 2 months ago:
It depends. “Donald Trump wants to annex Wisconsin as the 51st state” wouldn’t be funny. Pocketpair launching a Steam store page for the Palworld dating sim (which was last year’s April Fools joke) was.
But yeah, this isn’t the time for political humor.
- Comment on Hamster 2 months ago:
And the name of that hamster? Pikachu.
- Comment on 3's grip looks the most comfy 2 months ago:
Seconded. As far as pens are concerned, Uniball is where it’s at.
- Comment on People say they prefer stories written by humans over AI-generated works, yet new study suggests that’s not quite true 2 months ago:
I find that LLMs also tend to create very placative, kitschy content. Nuance is beyond them.
- Comment on Discord in Early Talks With Bankers for Potential I.P.O. 2 months ago:
Welp, there goes the neighborhood. If they want to do an IPO they’ll probably enshittify the hell out of the platform and jettison all remotely raunchy communities. Because nothing says “good investment” than a service that just drove out a fair chunk of its user base.
- Comment on Sergey Brin says AGI is within reach if Googlers work 60-hour weeks 2 months ago:
That undersells them slightly.
LLMs are powerful tools for generating text that looks like something. Need something rephrased in a different style? They’re good at that. Need something summarized? They can do that, too. Need a question answered? No can do.
LLMs can’t generate answers to questions. They can only generate text that looks like answers to questions. Often enough that answer is even correct, though usually suboptimal. But they’ll also happily generate complete bullshit answers and to them there’s no difference to a real answer.
They’re text transformers marketed as general problem solvers because a) the market for text transformers isn’t that big and b) general problem solvers is what AI researchers are always trying to create. They have their use cases but certainly not ones worth the kind of spending they get.
- Comment on People never understand the sacrifices I make for them. 3 months ago:
The prep and recovery blocks are also team calls; everyone prepares and recovers together, moderated by the scrum master.
- Comment on New Junior Developers Can’t Actually Code. 3 months ago:
Because giving answers is not a LLM’s job. A LLM’s job is to generate text that looks like an answer. And we then try to coax framework that into generating correct answers as often as possible, with mixed results.
- Comment on New Junior Developers Can’t Actually Code. 3 months ago:
I remember talking to someone about where LLMs are and aren’t useful. I pointed out that LLMs would be absolutely worthless for me as my work mostly consists of interacting with company-internal APIs, which the LLM obviously hasn’t been trained on.
The other person insisted that that is exactly what LLMs are great at. They wouldn’t explain how exactly the LLM was supposed to know how my company’s internal software, which is a trade secret, is structured.
But hey, I figured I’d give it a go. So I fired up a local Llama 3.1 instance and asked it how to set up a local copy of ASDIS, one such internal system (name and details changed to protect the innocent). And Llama did give me instructions… on how to write the American States Data Information System, a Python frontend for a single MySQL table containing basic information about the member states of the USA.
Oddly enough, that’s not what my company’s ASDIS is. It’s almost as if the LLM had no idea what I was talking about. Words fail to express my surprise at this turn of events.
- Comment on What can I actually do with 64 GB or RAM? 3 months ago:
Run a fairly large LLM on your CPU so you can get the finest of questionable problem solving at a speed fast enough to be workable but slow enough to be highly annoying.
This has the added benefit of filling dozens of gigabytes of storage that you probably didn’t know what to do with anyway.
- Comment on Anime Questions, Recommendations, and Discussion Thread [2025, Week 05] 4 months ago:
Yeah, Zenshu really delivered with the last episode. That was a level of ridiculousness I did not respect this show to reach.
- Comment on Weekly Recommendations Thread: What are you playing? 4 months ago:
System Shock (the remake) with a cut-down version of the Ironman mod to provide randomization. It’s only slight loot randomization so there’s no major pathing changes but it’s fun nonetheless.
I like randomizers. They add some additional replay value to already good games. I must’ve played through randomized Bloodstained a down times already – and twice that for Super Metroid. (And then there’s the beautiful mess that is randomized Borderlands 2. I don’t think I’m ever going to finish a run but man are they wild.)
- Comment on Technology Connections' thoughts on Mastodon 5 months ago:
Honestly, this suggests to me that the ability to defederate might be a bug rather than an issue.
If my instance doesn’t talk to the instance at foobar.example, I might be unable to see (parts of) relevant discussions. This is worse for a microblog like Mastodon than it is in the threadiverse but it’s still something to keep in mind even over here. And most non-enthusiasts don’t want to have to do that.
Email is an example of a successful federated platform and it barely has defederation support. But in general all mail servers can talk to all other mail servers as long as they provide the right look-at-me-I’m-legitimate signaling. That makes email easy to use for regular people no matter if they use Gmail or their cousin’s self-hosted mail server.
Perhaps that is how at least the non-threaded fediverse should work… However, that would also mean that some instance hosting heinous shit would keep being visible to everyone. It’s a tricky problem.
- Comment on Anime Questions, Recommendations, and Discussion Thread [2024, Week 48] 6 months ago:
Good season so far. I’m following four shows (as I usually do) and I’m happy that three of them are good to great.
Dan Da Dan is a rare instance of a show that can do everything. The characters are enjoyable, the comedy is on point, the action is punchy and easy to follow, and the show can even pull off drama really well when it wants to. Impressive. I don’t have much else to say, really.
Orb is the freshest work I’ve seen since Frieren. Possibly fresher than Frieren, actually.
There are virtually none of the staples ones would expect from anime; there is no world to be saved and society itself is more the villain than the literal torturer is. Stakes are personal but feel high. Action is very limited and it’s mostly a science-philosophical exploration of theological dogma through the lens of astronomy. It’s basically like a particularly talky Star Trek episode but in medieval Poland.
On top of that, the characters, presentation, and story are all of high quality.
I didn’t think the industry had something like this in it and it might beat Dan Da Dan as my anime of the season. While Dan Da Dan’s punches land flawlessly, Orb is a master of an entirely different martial art, one rarely seen in the genre.
Demon Lord 2099 is nothing particularly special but is a solid example of getting your balance right. The magical and cyberpunk aspects of the show are well-balanced, as are mild comedy, action, and drama. The show is aware that it can’t take itself too seriously. Honestly, the most jarring element is the random V-tuber cameo that fails to match the art style. All in all a good mix, although not one that I will really remember much about in a few years.
Mecha-Ude is a nice example of a show that doesn’t know where it’s strengths lie. The story is bland and clumsily told, the protagonist may as well not have bothered showing up, the action is stale, and the show is really bad at drama but thinks that’s one of its strengths. What it is good at is being silly – honestly, this could’ve worked well as a zany comedy. Its near-total lack of self-awareness keeps is from capitalizing on that, however.
- Comment on Absolute slander 6 months ago:
Oh, come on! The second picture shows that the three-packet hypothesis isn’t accurate either. It’s a 2.8 sauce packet cat.