teawrecks
@teawrecks@sopuli.xyz
- Comment on Brian Eno: “The biggest problem about AI is not intrinsic to AI. It’s to do with the fact that it’s owned by the same few people” 1 week ago:
Yes, and I don’t like the common comparison to binary blobs, and I’m attempting to explain why.
It is inherently safer to blindly run weights than it is to blindly execute a binary. The issues only arrise if you are then blindly trusting the outputs from the AI. But you should already have something in place to sanitize outputs and limit permissions, even for the most trustworthy weights.
It’s basically like hiring someone and wondering if they’re Hydra; no matter how deep your background check is, they could always decide to spontaneously defect and try to sabotage you. But that won’t matter if their decisions are always checked against enough other non-Hydra employees.
- Comment on Brian Eno: “The biggest problem about AI is not intrinsic to AI. It’s to do with the fact that it’s owned by the same few people” 1 week ago:
If you are familiar with the concept of an NP-complete problem, the weights are just one possible solution.
The Traveling Salesman Problem is probably the easiest analogy to make. It’s as though we’re all trying to find the shortest path through a bunch of points (ex. towns), and when someone says “here is a path that I think is pretty good”, that is analogous to sharing network weighs for an AI. We can then all openly test that solution against other solutions and determine which is “best”.
What they aren’t telling you is whether people traveling that path somehow benefits them (maybe they own all the gas stations on that path. Or maybe they’ve hired highway men to rob people on that path). And figuring out if that’s the case in a hyper-dimensional space is non-trivial.
- Comment on Tesla Stock Is Plunging Again. It Could Drop for a Ninth Straight Week. 2 weeks ago:
And then the citizens will own a chunk of Tesla? And we will see our investment pay off when it does well?
Pretend I made that into padme/anakin meme.
- Comment on The most influential video game of all time - BAFTA 3 weeks ago:
These were the first two to come to mind for me as well. I hate what they’ve become, especially wow, but they were both clearly extremely influential.
- Comment on Firefox deletes promise to never sell personal data, asks users not to panic 4 weeks ago:
It seems like the issue here is, users want to be spoken to in colloquial language they understand, but any document a legal entity produces MUST be in unambiguous “legal” language.
So unless there’s a way to write a separate “unofficial FAQ” with what they want to say, they are limited to what they legally have to say.
And maybe that’s a good thing. Maybe now they need to create a formal document specifying in the best legalese exactly what they mean when they say they “will never sell your data”, because if there’s any ambiguity around it, then customers deserve for them to disambiguate. Unfortunately, it’s probably not going read as quick and catchy as an ambiguous statement.
- Comment on Bill proposed to outlaw downloading Chinese AI models. 1 month ago:
I agree that you can’t know if the AI has been deliberately trained to act nefarious given the right circumstances. But I maintain that it’s (currently) impossible to know if any AI had been inadvertently trained to do the same. So the security implications are no different. If you’ve given an AI the ability to exfiltrating data without any oversight, you’ve already messed up, no matter whether you’re using a single AI you trained yourself, a black box full of experts, or deepseek directly.
But all this is about whether merely sharing weights is “open source”, and you’ve convinced me that it’s not. There needs to be a classification, similar to “source available”; this would be like “weights available”.
- Comment on Bill proposed to outlaw downloading Chinese AI models. 1 month ago:
Those security concerns seem completely unrelated to the model, though. You can have a completely open source model that fits all those requirements, and still give it too much unfettered access to important resources with no way of actually knowing what it will do until it tries.
- Comment on Bill proposed to outlaw downloading Chinese AI models. 1 month ago:
Is there any good LLM that fits this definition of open source, then? I thought the “training data” for good AI was always just: the entire internet, and they were all ethically dubious that way.
What is the concern with only having weights? It’s not abritrary code exectution, so there’s no security risk or lack of computing control that are the usual goals of open source in the first place.
To me the weights are less of a “blob” and more like an approximate solution to an NP-hard problem. Training is traversing the search space, and sharing a model is just saying “hey, this point looks useful, others should check it out”. But maybe that is a blob, since I don’t know how they got there.
- Comment on Bill proposed to outlaw downloading Chinese AI models. 1 month ago:
It’s also the free market for those corporations to buy a government and use it to outlaw competition.
- Comment on China's new and cheaper magic beans shock America's unprepared magic bean salesmen 1 month ago:
Yeah, I agree that in the long term those two sentiments are inconsistent, but in the short term we have to deal with allegedly misguided layoffs, and worse user experiences, which I think makes both fair to criticise. Maybe firing everyone and using slop AI will make your company go bankrupt in a few years, and that’s great; in the meantime, employees everywhere can rightfully complain about the slop and the jobs.
But yeah, I don’t think it’s fair to complain about how “inefficient” an early technology is and also call it “magic beans”.
- Comment on China's new and cheaper magic beans shock America's unprepared magic bean salesmen 1 month ago:
Hah, see that’s what I thought when various family members asked if I had heard about it. Turns out, if our electronics need grounding, so must our bodies…
- Comment on China's new and cheaper magic beans shock America's unprepared magic bean salesmen 1 month ago:
I have made only factual statements. You can believe I’m arrogant for doing so, you can believe the preference of hundreds of millions of people is “niche” or “few” in number. Those are called opinions.
Which statements have I made that you believe to be my opinion?
- Comment on China's new and cheaper magic beans shock America's unprepared magic bean salesmen 1 month ago:
Yeah, I understand that you personally choose to disagree with reality, maybe you don’t like what reality has become, but unfortunately that doesn’t make it less real.
Twitter wasn’t profitable for its entire existence, it’s often a cesspool of ragebaiters, but clearly it has value because the second it was taken over, everyone insisted on continuing to use it, even choosing to migrate to various clones.
Uber and Lyft have been struggling to be profitable by effectively stealing from their drivers, but millions of people get off a plane and immediately use the services every day. It clearly has value.
Same for doordash and uber eats.
Your personal distaste for the business practices are valid, but they’re not relevant when discussing what the current state of the technology is. For many millions of people, chatgpt has (for better and worse) replaced traditional search engines. Something like 80% of students now regularly use AI for their homework. When Deepseek released, it immediately jumped to #1 on the Apple Store.
None of that is because they’re “magic beans” from which no value sprouts. Like it or not, people use AI all. the. time. for everything they can imagine. It objectively, undeniably has value. You can staunchly say pretend it doesn’t, but only if you are willingly blind to the voluntary usage patterns of hundreds of millions (possibly billions) of people every hour of every day.
And for the record, I am not in that group. I do not use any LLMs for anything currently, and if anything makes me use AI against my will, I will promptly uninstall it (pun intended).
- Comment on China's new and cheaper magic beans shock America's unprepared magic bean salesmen 1 month ago:
No opinions whatsoever. I believe I made that clear in my list of things to disregard when considering the objective reality of current AI tech.
- Comment on China's new and cheaper magic beans shock America's unprepared magic bean salesmen 1 month ago:
Lol this article is very relevant to a lot of scam industries (essential oils, Earthing, 5G protection crystals, etc), but AI is objectively not one of them.
Regardless of how much of a bubble we’re in, regardless of how many bad ideas are being pushed to get VC funding or pump a stock, regardless of how unethical or distopian the tech is, AI objectively has value. It’s proving to be the most disruptive tech since the world wide web (which famously had a very similar bubble of bad ideas), so to call it “magic beans” is just wishful thinking at best.
- Comment on Facebook Says It’s Not Forcing You to Follow Trump 2 months ago:
“Facebook says it’s not forcing you to use Facebook”
- Comment on "Meta and X are going rogue:" European Digital Rights group (EDRi) urges EU to invest in infrastructure "like Mastodon, Peertube and other key pieces of the Fediverse" to secure Europe's independence 2 months ago:
The US leadership right now, maybe, but remember that Trump didn’t win so much as the incumbent lost. Most Americans didn’t vote for right-wing policies, they voted against inflation and housing costs. It’ll only take a year or two before people start realizing that Trump can’t fix the problems either (or won’t, because that would mean eating the rich).
So yeah, probably no alliance in the short term, but the US isn’t even its own ally right now, so we need to see how this all shakes out before we know how we’ll align with the EU in the long term (i.e. beyond this term).
Trump knows this, and he’s also been advised that the one thing that historically restores popularity for a leader is expanding a country’s territory. So my guess is that, the worse Trump’s approval rating is, the more likely it’ll be that he tries to take Greenland or Panama. Which I think is still a huge gamble for his approval rating.
- Comment on I love my smart TV (From Mastodon) - Repost 2 months ago:
Yeah, i don’t like that all the more recent devices all added remotes and explicit apps you have to install and launch. Also the “newest” 4k-capable chromecast is from 2020, so I would already be upgrading to old hw that’s a worse experience.
- Comment on I love my smart TV (From Mastodon) - Repost 2 months ago:
Hah, I was going to say, I do check for updates at least once when I first get it, because I have run into TVs that shipped with HDR bugs in the stock firmware.
For the Chromecast, what happens with yours? Mine randomly restarts, or reconnects to wifi, or sometimes Plex has trouble buffering until I reboot it.
I recently bought a raspi5 to try out FCast, though currently afaik only Grayjay supports it.
- Comment on I love my smart TV (From Mastodon) - Repost 2 months ago:
I just never connect my TV to the internet and never have any problems. My old Chromecast is showing its age though.
- Comment on This Year, RISC-V Laptops Arrive 2 months ago:
I assume DeepComputing isn’t releasing any of their designs as open source, right? They’re just producing RISC-V compatible chips?
- Comment on Technologist: 'Fining Big Tech isn't working, make them give away illegally trained LLMs as public domain' 3 months ago:
What the? I’m literally saying what action to take, what is happening? Is there maybe a bug where you only see the first few characters of my post? Are you able to read these characters I’m typing? Testing testing testing. Let me know how far you get. Maybe there’s just too many words for you? Test test. Say “elephant” if you can read this.
- Comment on Technologist: 'Fining Big Tech isn't working, make them give away illegally trained LLMs as public domain' 3 months ago:
we will not land in a society where the general public profits from not having work. It will be the same owners of capital profiting as per usual.
If we do nothing, sure. I’m suggesting, like the article, that we do something.
The only sentiment I took issue with was the poster above who suggested that somehow the solution would be to delete/destroy illegally trained networks. I’m just saying that’s not practical nor progressive. AI is here to stay, we just need to create legislature that ensures it works for us, especially when it couldn’t have been built without us.
- Comment on Technologist: 'Fining Big Tech isn't working, make them give away illegally trained LLMs as public domain' 3 months ago:
I didn’t misinterpret what you were saying, everything I said applies to the specific case you lay out. If illegal networks were somehow entirely destroyed, someone would just make them again. That’s my point, there’s no way around that, there’s just holding people accountable when they do it. IMO that takes the form of restitutions to the people proportional to profits.
- Comment on Technologist: 'Fining Big Tech isn't working, make them give away illegally trained LLMs as public domain' 3 months ago:
For sure, here you go.
- Comment on Technologist: 'Fining Big Tech isn't working, make them give away illegally trained LLMs as public domain' 3 months ago:
I understand that you are familiar with the buzzword “LLM”, but let me introduce you to a different one: transformers.
Virtually all modern successful AIs are based on transformers, LLMs included. I agree that LLMs currently amount to a chinese-room-inspired parlor trick, but the money involved has no doubt advanced all transfomer-based AI research, both directly (what works for LLMs may generalize) and indirectly (the market demand for LLMs in consumer products has created the a demand for power and compute hardware).
We have transformer-based AI to thank for our understanding of the covid19 protein, and developing a safe and effective vaccine in a timely manner.
The massive demand for energy has convinced Microsoft, Meta, and others to invest in their own modern nuclear power plants, representing a monumental step forward in sustainable energy generation that we have been trying to convince the US government to take for decades.
Modern AI is being used to solve the hardest problems of nuclear fusion. If we can finally crack that nut, there’s no telling what’s possible.
But specifically when it comes to LLMs, profitable or not, people obviously find them useful. People aren’t using it in place of search engines, or doing all their homework with it because they don’t find it useful. My only argument is that any AI trained on public content without consent should be required to effectively buy a license from, or pay royalties to the public. If McDonald’s is going to replace their front counters with AI trained on public content, then they should have to pay taxes proportional to how much use they get from that AI.
In the theoretical extreme, if someone trains an AI on the general public’s data, and is able to create an AI that somehow replaces every job on earth, then congrats, we now live in a post-work society, we just need to reach out and take it rather than letting one person capitalize infinitely.
And at the end of the day, if you honestly believe the profits from AI are non-existent, then what are you worried about? All those companies putting all their eggs in the LLM basket are going to disappear overnight when the AI bubble finally pops, right?
- Comment on Technologist: 'Fining Big Tech isn't working, make them give away illegally trained LLMs as public domain' 3 months ago:
Destroying it is both not an option, and an objectively regressive suggestion to even make.
Destruction isn’t possible because even if you deleted every bit of information from every hard drive in the world, now that we know it’s possible, someone would recreate it all in a matter of months.
Regressive because you’re literally suggesting that we destroy a new technology because we’re afraid of what it will do to the technology it replaces. Meanwhile, there’s a very decent chance that AI is our best chance at solving the energy/climate crises through advancing nuclear tech, as well as surviving the next pandemic via ground breaking protein folding tech.
I realize AI tech makes people uncomfortable (for…so many reasons), but becoming old fashioned conservatives in response is not a solution.
- Comment on Blacksky Is Nothing Like Black Twitter—and It Doesn’t Need to Be 3 months ago:
The problem in many cases isn’t that they don’t literally see it but that they aren’t aware of what constitutes racism a lot of the time.
I agree with this part, “in many cases” sure,
That’s the primary issue here.
…but I think this a strong claim to make unless you have data to back it up.
I believe you and I are likely speaking from our own anecdotal experience on the platform, and for all we know, most people are in instance bubbles and are also speaking from their own perspectives.
If the “primary issue” is “why do some people not report seeing racism?” and the two possible explanations are either “they see it but are not aware” and “they actually never see it”, then unless we have accurate data from all those bubbles, we can’t make any claims about which is the real explanation.
But if you have data on this, that would change everything.
- Comment on Blacksky Is Nothing Like Black Twitter—and It Doesn’t Need to Be 3 months ago:
Comparing the “racism” present on a federated service to that on a centralized one doesn’t make sense. You can say certain instances of the service fail to adequately moderate racism, but there are so many niche pockets of mastodon that most people are exposed to, and moderated by, completely different groups.
To make a slightly more nerdy analogy, it’s like someone saying “the windows desktop experience is better than Linux”. Well Linux doesn’t come with a desktop interface, so that statement doesn’t make sense. Which of the dozens of windowers/distros are you talking about? I’m sure the criticism is fair, but it doesn’t contain enough information to make any real claim.
So it’s not unreasonable for one person to say “I see racism on Mastodon” and many others to say “I never see it”, and not just because of the races of the people involved. “Mastodon” refers to a protocol, not the various ecosystems that use it.
- Comment on Silent but Deadly: I met some of my closest friends through multiplayer games. Then a strange happening turned everyone (literally) speechless. 3 months ago:
“Given the opportunity, players will optimize the fun out of a game.”