Post B5Fb1KFdoC2uB5eCbg by burnoutqueen@todon.nl
 (DIR) More posts by burnoutqueen@todon.nl
 (DIR) Post #B5FaSXUcaJNR6v4CzQ by libreleah@mas.to
       0 likes, 0 repeats
       
       fun fact: AI is a threat to democracy everywhere. the companies that control it have the power to literally control what humans see and think. it's being embedded everywhere - in search engines, in operating systems, the NEWS, the courts...fun fact: democracy is good. AI is... not bad, just needs to be not controlled by like, 3 giant corporations. maybe democratise it, make it open source, with training fully transparent and done in a distributed way (publicly administrated).nationalised AI?
       
 (DIR) Post #B5Fb1KFdoC2uB5eCbg by burnoutqueen@todon.nl
       0 likes, 0 repeats
       
       @libreleah ahhh yes, let's nationalize the torment nexus
       
 (DIR) Post #B5Fb38jHcMIz1SNSMK by libreleah@mas.to
       0 likes, 0 repeats
       
       lots of things are good or bad depending on control. for example:if the media is owned and controlled by decree of a small clique of people, with information routinely censored or skewed, that is what you see in dictatorships, and the media then is actively harmful to the human mind.but we value a free press.*computers* can do harm, if not under your control. but we have Free Software, and then your computer is your best friend ever.AI is the same. we should control it democratically.
       
 (DIR) Post #B5FbjB8D4qgSdQ5A12 by libreleah@mas.to
       0 likes, 0 repeats
       
       i see a handful of companies controlling it, and using it for bad. if *we* controlled it, democratically, we could use it to liberate humanity, freeing it from tedious drudgery.when used responsibly, it can make you more productive. there are hospitals now trialling it in surgeries, and general care. numerous industries are using it to become more efficient.a democratically controlled AI could be used to improve human quality of life. we do not yet have AI democracy; we have AI dictatorship.
       
 (DIR) Post #B5Fh29hX7PsxTJTMkS by rl_dane@polymaths.social
       0 likes, 0 repeats
       
       @libreleahValid points, but we cannot ignore how an "AI" procures its training data, or how it's run (energy, pollution, water consumption, etc.)
       
 (DIR) Post #B5Fh29viGhECBI6gqm by libreleah@mas.to
       0 likes, 0 repeats
       
       @rl_dane we can handle 20x as much electricity usage worldwide, if we would invest properly in renewable infrastructure. solar, wind, hydroelectric. and in the future, maybe fusion will be viable at commercial scale.water usage is still a problem (cooling and such), but there must be a way to cool the datacentres sustainably. and a more distributed AI infrastructure would put little strain on baseload energy supply, especially if home/business has solar and wind (off-grid electricity).
       
 (DIR) Post #B5FhYMUT6V18P5id8a by rl_dane@polymaths.social
       0 likes, 0 repeats
       
       @libreleahI think one of the big problems is that is has been a solution in search of a problem.Using language models in things like translation, or in medical applications where they actually help and don't just belch out nonsense, I can see a case for. But for the average person? I honestly don't think the average person should interact with an AI. It's very harmful for intellectual development.Also, I loathe the fact that major search engines have replaced the first screenful of search with AI hallucinations. That's not only wasteful of energy, it's fundamentally a lie, and theft.
       
 (DIR) Post #B5FhYMnbxKKVMSfuyW by libreleah@mas.to
       0 likes, 0 repeats
       
       @rl_dane yeah that pisses me off too. i'm trying to google something and the first half of the result page is AI. i referred earlier to "responsible use" of AI; your own example is a perfect demonstration of irresponsible use. shoving it in everyone's face by default is just irresponsible. if someone wants to use it, they can always just open a chat window
       
 (DIR) Post #B5FifuEc8qRz2bSr56 by rl_dane@polymaths.social
       0 likes, 0 repeats
       
       @libreleahAs it is, people use ChatGPT way too much, and it freaking makes me ill.We finally got to the point where a $10 SBC could serve the computing needs for like 100 people, and they want to go back to ENIAC-level energy expenditure.This whole thing is stupid, and I'm not at all above throwing out the baby with the bathwater when the bathwater came from inside Chernobyl. :PThe best I've heard is vague promises of how LLMs can help cancer research, but all I've seen for concrete examples so far is utterly demented and pointless. Stories like a woman at her doctor's office getting stares because the computer flagged her as trans (yes, of course, the fact that she's getting ugly stares because they think she's trans is bad enough) because the doctor apparently dictated "the patient is a female mailman" into the "AI"-enhanced dictation system.This all reeks of the 2001 dotcom bubble, but absolutely 10x worse.It's one of the biggest disruptors in human history with practically no net benefit, except to dirtbags like Sam Altman and Peter Thiel.
       
 (DIR) Post #B5FifurxmXvT0dsHPk by libreleah@mas.to
       0 likes, 0 repeats
       
       @rl_dane i'm 100% against the current monopolised iteration of AI, just because we have no control of it. we need something like the free software charter, for AI, and a well-established and distributed infrastructure for training it. in this context, i mean that anyone who wants to could contribute some of their computing capacity to train models, and this would become the norm worldwide. using some proof of work scheme (read: not blockchain).we don't have that yet. it's all centralised.
       
 (DIR) Post #B5Fj9MGX0CxeNKPAa8 by libreleah@mas.to
       0 likes, 0 repeats
       
       @rl_dane and there ought to be a new law: crime against computing. the recent memory price hikes as a result of openai's treachery would fall under this new law. fines would be imposed :)or maybe just actually enforce competition law. it's a thing. it is not always enforced. but it can be enforced, with positive results. pretty much all of the AI companies are engaging in anti-competitive behaviour, endorsed by their respective governments.
       
 (DIR) Post #B5FjZpuJPyprpudbuq by rl_dane@polymaths.social
       0 likes, 0 repeats
       
       @libreleahAnd most of the "Open Source A.I." definitions say nothing about how the training data was sourced, so it still has huge ethical issues.But fundamentally, I don't want to control it, I want it to go away, for the most part. I know that's not realistic, but I'm kinda tired of dancing with the devil in the name of pragmatism. XD
       
 (DIR) Post #B5FjZq6MhATcRIHEhc by libreleah@mas.to
       0 likes, 0 repeats
       
       @rl_dane that genie is out of the bottle now. it's never going back in.
       
 (DIR) Post #B5FjsVcH5ETEPJ6YZU by rl_dane@polymaths.social
       0 likes, 0 repeats
       
       @libreleahIf #TheGoodPlace has taught me anything, there's nothing that can't be fixed with a good molotov cocktail. /j
       
 (DIR) Post #B5FjsVlqVe7uszaCUS by libreleah@mas.to
       0 likes, 0 repeats
       
       @rl_dane and a dubious dance crew from florida