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 so, I find this discussion disappointing for a few reasons.
The biggest one is this: all three platforms that @carnage4life calls out are connected via ActivityPub. They are on one inter-network.
In theory, he should not need three different accounts, with three different follower groups. He should have one account, and all 103k followers (minus duplicates!) could be part of the same conversation, on whatever server platform they use.
In practice, few people do this today.
As technologists we need to do more to smooth those junctures and make them less of a barrier. I hope in a few years when @carnage4life looks at his network, it feels more integrated and less separated.
@evan @panos @scottjenson @carnage4life Two things can be true at the same time:
- Using LLMs is unacceptable
- We should not abuse individuals for using LLMs, both because that is ineffective at stopping LLM use, and because everyone does harmful things to survive capitalism
The real problem is when people deny the harms, or decide to ignore the harms because of "inevitability" etc. It's understandable that you want the thing you're doing to not be harmful, but wishing won't make it so.
@panos @evan @carnage4life @scottjenson This is what I'm talking about. You're both minimizing/denying harms and saying they don't matter. This is one of the biggest problems with LLMs, they turn people into apologists for the fossil fuel industry because they don't want to think they're helping destroy the world.
*If* flying is more harmful, that's no excuse. There's always something more harmful until you reach the top, and then the excuse will be it's too important or too difficult to stop.
@skyfaller if flying is more harmful, and one still flies, then there is no convincing argument that a behaviour less harmful by orders or magnitude is absolutely unacceptable.
@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.
Thanks @scottjenson for the props. Just a quick correction: #Apertus was trained by a research consortium on publicly funded infrastructure, but is not government-run.
Let me know if you're interested in knowing more about that 'bathwater'.
is there a conceptual analysis of factors behind energy costs for developing LLM's. E.g., as function of "size" (working assumption that this a relevant metric of "usability" 🤣 )
I am curious to what degree there is a measurable impact from "urgency" to iterate fast, capture markets, IPO or whatever, versus slower (and lower energy?) hardware options.
Scott Jenson
Evan Prodromou
Nelson
panos 
o lаvrоvsky
Open Risk