Is #mastodon becoming an echo chamber?
-
@evan @panos @carnage4life @scottjenson I do not concede that LLMs are "orders of magnitude" less harmful than flying. Also I do not fly.
Anyone dismissing LLM harms doesn't understand the scale of the climate crisis or of LLMs. Sadly, “It is difficult to get a man to understand something, when his salary depends upon his not understanding it!” I can't force this understanding onto anyone in a few toots, they would have to want to understand, when the industry requires LLMs if you want to eat.
@skyfaller @evan @panos
Don't throw the baby out with the bathwater. There are ethically trained models (apertus from the goverment of Switzerland comes to mind) as well as many open source small language models that run locally and do not burn down the planet.I'm with you, the large foundational models are a crime and horrible and shouldn't be used. But that doesn't mean we can't discuss local, open, and ethical models.
-
@scottjenson @carnage4life on the topic of AI, I find the abusive conversations on the Fediverse pretty dispiriting. People I like and respect have worked themselves into the position that use of AI is an inexcusable sin, and that anyone who uses AI merits harassment and abuse. Given that 85% of developers use or plan to use AI (Stack Overflow poll), that means a huge number of tech people getting brigaded by our anti-AI squad.
@evan @scottjenson @carnage4life I agree. I would rather talk about how we can improve LLMs and their applications than post anti-AI memes and shame people who use LLMs.
For example, let's use more voluntary training data, let's make smaller, more efficient models, let's do more quality control with the output, let's protect authors and artists from having their work stolen, let's not over-rely on LLMs or use them for things they are bad at. These are actionable steps we can take to improve the world with LLMs in it.
I do not believe that the "LLMs are categorically evil" approach is going to have any good results. The genie is out of the bottle, people find this technology very useful in certain ways. We might as well try to reduce the harms and improve the outcomes of using LLMs rather than chase after a cultural or legal prohibition which will never really be effective.
-
@stairjoke @thibaultamartin I'm happy to get comments like this. It's bit frustrating that there do appear to be two camps here, although I'm not surprised.
Part of the reason I "bait" these conversations is to at least bring this discussion into the open. We must realize that our culture *is* the driving force behind our success (or failure)
@scottjenson @thibaultamartin I have to admit, I’m quite tired of social media. I made an account when MySpace was popular ca. 2005? I’ve been online in one way or another ever since.
The most valuable things online for me were YouTube starting in 2006 and early day Podcasts, which I downloaded to my iPod starting in 2004. On third place, and with considerable margin are blogs (I wrote one 2006 – 2010, I’m back at it since 2022).
No medium moving faster than blogs has truly benefited me much.
-
@scottjenson @carnage4life @evan I think step 1 is not calling all of it AI, an extremely broad vague term makes it very hard to have a nuanced discussion. If Mozilla uses a local ML model to detect which field on the page is which type to autofill better, that's very different from a remote LLM chat bot.
@Aurimas One part is not different: knowing how and from which data were these model trained. Claiming "it's fine it's a local ML model" is far from making the solution an ethical one.
-
Is #mastodon becoming an echo chamber? This post from @carnage4life has me questioning our community. The Mastodon team is finally getting some traction, the product improvements are increasing, The #UX is improving, yet people posting on multiple platforms are making comments like this. It's confusing.
I *know* people here don't want this to be a classic social media-clone but we'd *like* journalists to be here right? They aren't coming with examples like this!
@scottjenson i think it's a bigger problem than most people realize (or care). i joined in 2017. at that time it felt like everyone was excited when one more person came on board. i feel the vibe now is for walls to keep people out. (to be clear, i'm not talking about mastodon specifically—i'm talking about the fediverse more generally.) my feeling about networks is you grow if you accept diversity and are willing to support multiple communities, or you shrink down to a monoculture.
-
@evan Fwiw, examples like that are one of the main reasons why I'm pretty quiet on this platform. I love the idea of ambiently sharing things with friends and peers, but I'm just not ready to deal with something like that.
@wlach I'm sorry to hear that.
-
@scottjenson @thibaultamartin I have to admit, I’m quite tired of social media. I made an account when MySpace was popular ca. 2005? I’ve been online in one way or another ever since.
The most valuable things online for me were YouTube starting in 2006 and early day Podcasts, which I downloaded to my iPod starting in 2004. On third place, and with considerable margin are blogs (I wrote one 2006 – 2010, I’m back at it since 2022).
No medium moving faster than blogs has truly benefited me much.
@scottjenson @thibaultamartin I often feel like the time I spent on social media is wasted, because it doesn’t lead to anything.
Worse, actually, instead of building something lasting, I spend time on socials. It’s extremely hard to resist, but I’m working on a blog system easier to post to than to socials.
What really irks me is that here in Germany, basically all events, from protests to breakfast with friends, is organised via Instagram or WhatsApp.
-
@evan @panos @carnage4life @scottjenson I do not concede that LLMs are "orders of magnitude" less harmful than flying. Also I do not fly.
Anyone dismissing LLM harms doesn't understand the scale of the climate crisis or of LLMs. Sadly, “It is difficult to get a man to understand something, when his salary depends upon his not understanding it!” I can't force this understanding onto anyone in a few toots, they would have to want to understand, when the industry requires LLMs if you want to eat.
@skyfaller@jawns.club I believe that (further) ecological descruction is sadly unavoidable under capitalism. No matter how more ecologically some of us try to live, those who profit from this will only see any climate gains from our behavior as a chance to do even more damage on their own. I don't mean that this means that everything is acceptable, I also do what I can from other aspects, I just don't believe that I am responsible for the climate crisis and that I/we could avert it, sorry, you can't guilt trip me into this, that's victim blaming and thanks but no. @evan@cosocial.ca @carnage4life@mas.to @scottjenson@social.coop
-
@rakoo oh FFS put your pitchfork down. You're creating a strawman and claiming it's me. I'm not equating the two! I'm saying "intellectual purity" tests cut both ways. If you're willing to slam all "AI people" as techbros influencers and want to kick them out, that can slide into any other topic or group.
My goal was to say inclusivity is hard, *especially* when it concerns groups that aren't marginalized. But not caring about these people makes it even easier to then affect marginalized peope.
I'm bringing this topic up BECAUSE black twitter was chased off this platform in 2022 and I'm really pissed that we haven't learned our lesson, we're still chasing people off (even if they aren't marginalized)
@scottjenson The reasons black people were rejected are _not_ the same reasons techbros are being rejected. You need to learn about what racism is: it's not a lack of inclusivity or being closed to different opinions, it's the structure we are all swimming in that is racist and needs to be worked on. It's the black people saying "please don't make it hard for us" and not listening to them. The women saying "replyguys are annoying" and not giving them an infrastructure that works for them by default. The work is on acknowledging the structure we're in, the actions we do that keep it in motion.
You're approaching this as a simple intellectual dissensus, a divergence in opinions, when it just isn't.
Groups that aren't marginalized don't matter. Society gives them all the tools, comfort and mindspace to take care of what they want to do. Focusing on them only enforces the marginalization of others. Instead, focusing on the marginalized and their needs help reduce disparities for everyone instead.
Non-marginalized people didn't leave because they were suffering yet again the same marginalization they endure everywhere else; they leave because they don't find what they want. Which is fine. -
@Aurimas One part is not different: knowing how and from which data were these model trained. Claiming "it's fine it's a local ML model" is far from making the solution an ethical one.
@fabrice hence the need for precise terms
-
@scottjenson The reasons black people were rejected are _not_ the same reasons techbros are being rejected. You need to learn about what racism is: it's not a lack of inclusivity or being closed to different opinions, it's the structure we are all swimming in that is racist and needs to be worked on. It's the black people saying "please don't make it hard for us" and not listening to them. The women saying "replyguys are annoying" and not giving them an infrastructure that works for them by default. The work is on acknowledging the structure we're in, the actions we do that keep it in motion.
You're approaching this as a simple intellectual dissensus, a divergence in opinions, when it just isn't.
Groups that aren't marginalized don't matter. Society gives them all the tools, comfort and mindspace to take care of what they want to do. Focusing on them only enforces the marginalization of others. Instead, focusing on the marginalized and their needs help reduce disparities for everyone instead.
Non-marginalized people didn't leave because they were suffering yet again the same marginalization they endure everywhere else; they leave because they don't find what they want. Which is fine.@rakoo That was an amazing answer. Thank you.
-
@evan @scottjenson @carnage4life I agree. I would rather talk about how we can improve LLMs and their applications than post anti-AI memes and shame people who use LLMs.
For example, let's use more voluntary training data, let's make smaller, more efficient models, let's do more quality control with the output, let's protect authors and artists from having their work stolen, let's not over-rely on LLMs or use them for things they are bad at. These are actionable steps we can take to improve the world with LLMs in it.
I do not believe that the "LLMs are categorically evil" approach is going to have any good results. The genie is out of the bottle, people find this technology very useful in certain ways. We might as well try to reduce the harms and improve the outcomes of using LLMs rather than chase after a cultural or legal prohibition which will never really be effective.
@earth_walker @scottjenson @carnage4life
One thing we don't talk about, when we talk about AI, is that, for hackers, AI-assisted software development threatens our livelihoods and lifestyle. It undermines the special position that we hold in the social and economic order.
No amount of lowering power consumption, careful training data provenance, or decentralised deployment will help with that.
-
@earth_walker @scottjenson @carnage4life
One thing we don't talk about, when we talk about AI, is that, for hackers, AI-assisted software development threatens our livelihoods and lifestyle. It undermines the special position that we hold in the social and economic order.
No amount of lowering power consumption, careful training data provenance, or decentralised deployment will help with that.
@earth_walker @scottjenson @carnage4life
It would be interesting to have the discussion of how, if we don't manage to abolish all LLM-assisted software development entirely, we can maintain hacker culture and a positive influence on the world's use of technology.
-
@scottjenson @Gargron “AI people” are not a protected class. It seems much more important that we focus on being a welcoming and inclusive platform for protected classes, particularly actual marginalized communities.
@dgodon @scottjenson @Gargron consider though that there are people in marginalized communities who are also "AI people". are they not welcome? is anti-AI (or anti-whatever) more important than welcoming marginalized people? literally, where is the line?
-
@evan @panos @carnage4life @scottjenson I do not concede that LLMs are "orders of magnitude" less harmful than flying. Also I do not fly.
Anyone dismissing LLM harms doesn't understand the scale of the climate crisis or of LLMs. Sadly, “It is difficult to get a man to understand something, when his salary depends upon his not understanding it!” I can't force this understanding onto anyone in a few toots, they would have to want to understand, when the industry requires LLMs if you want to eat.
@skyfaller My work has been up until recently building software to make greenhouse gas inventories. I am well aware of the causes of the climate crisis, and I can tell you categorically that AI is not a significant one. You've already seen my math on the topic, but I can share the links again if you need them.
-
@evan @scottjenson @carnage4life sure, let’s take steps to prevent abuse and make this a more welcoming and inclusive space, but let’s stop pretending that “AI users” are a marginalized community. It’s like arguing that cops or Republicans are a protected class. Center actual marginalized groups in these discussions! If they feel welcome then there’s a better chance non-hype AI users will too
@dgodon your argument is, i guess, that there are no members of marginalized communities who are "AI users"? or is it that it doesn't matter because your being against AI and their using AI is more important than anything else about who they are?
-
@rakoo That was an amazing answer. Thank you.
@scottjenson you might not be the stereotypical pro-white pro-cis man, but the focus of your attention and the way you approach systemic issues might put forward those people: we can't naively say "I'm clean", we're all a part of this because we've all been in there for so long. I think the very first move is to listen to what marginalized people have been saying for ages and start from there. There is no need for yet another discussion, as if the topic was new or still too vague. Recognizing marginalized people, considering them as equals whose voice and expertise should guide us, this is where we must start -
@skyfaller My work has been up until recently building software to make greenhouse gas inventories. I am well aware of the causes of the climate crisis, and I can tell you categorically that AI is not a significant one. You've already seen my math on the topic, but I can share the links again if you need them.
@skyfaller AI is projected to rise to as much as 1/3 of all IT emissions by 2030, so about 0.3% of global emissions. Air travel is about 3.5% of global emissions. That's an order of magnitude.
-
@skyfaller AI is projected to rise to as much as 1/3 of all IT emissions by 2030, so about 0.3% of global emissions. Air travel is about 3.5% of global emissions. That's an order of magnitude.
@skyfaller for individuals, an hour of flight can emit about 1kg CO2. An hour of LLM use on a dirty grid emits 0.01kg of CO2.
-
@earth_walker @scottjenson @carnage4life
It would be interesting to have the discussion of how, if we don't manage to abolish all LLM-assisted software development entirely, we can maintain hacker culture and a positive influence on the world's use of technology.
@evan @scottjenson @carnage4life I would argue that the history of technology is defined by working with increasing levels of abstraction. First you were plugging in wires, then you had simple instruction sets, then low level languages, then high level languages, and now we can use natural language to write software. Every time this happened, we found new sources of inspiration and made cool and useful new things. I see LLMs as part of that story and not fundamentally different. In my opinion, hackers are ultimately people who trade in ideas, the technology is more the means to actualize the ideas. If you get too attached to specific technologies you'll have a problem when the world changes and the focus shifts to new technologies. So I see the cultural side of the issue as something that people can potentially adapt to.
That said, yes LLMs being pushed by capitalist entities are definitely reducing the economic value of information-based labor. But that's unfortunately also the latest iteration of a long story of industrialization and automation. I believe we should fight against the devaluation of labor by capitalists, but I think that we should be more focused on policy than technology in that fight.
