This type of purity politics does no one any favors and frankly is just fucking annoying. What a waste of time and attention this afternoon. No one with a passing familiarity with @elementary would think they are a legitimately pro AI organization and yet!
As much as I am against AI, I oftentimes find the purity battle tedious. It's a losing game to try and detect LLM usage by community members. We need to be more focused on setting acceptable boundaries in our communities and welcoming back those people who are tired of LLMs and AIs or who can't escape them because some reason. Squabbling over the last 5% of those boundaries will not help us beat back the horde of AI apolgia and money.
This is one of the most moving stories I've read this year in journalism. I hope that this family is reunited and that more families in the future are not destroyed arbitrarily by harmful policy.
Some weeks ago I read an excellent blog post about community building, LLMs, and Open Source that I really wish I could find again. The quick summary was that communities grow by welcoming in members deeper into them as they participate and that this was how they grow and replenish themselves. LLMs cut people off from the community because they intercept them early on. This is the loss feel twice. Once inside the community and once outside it
Do you use a file manager on your phone?
Tell me the Linux Foundation is structured to support corporate entities without telling me it's structured to support corporate entities.
In a lot of ways, this is the ideal way to use a computer. No esoteric commands, just instructions. My colleagues don't want to worry about a tool (and especially the power dynamics behind it!) they want to take the text out of a document and transform it for someone else. LLMs, especially in their subsidized form, represent an extremely efficient way of doing this. Handwritten meeting notes OCR'd into a summary to an email? Claude can do that no problem. It's a single tool.
One of the things I'm noticing in the last several months seeing folks interact with LLMs (and reading the providers own docs) is how much they remind me of Unix piping at a natural language level. As looping and tool calls become more prevalent this seems to be especially true. The workflow becomes: take input, execute action react, check result repeat as needed. It's an especially powerful feeling and use case! And it's all in relatively plain everyday language (WTF is a sed anyways???)
It's disappointing to me to find out that LLMs seem to have become the defacto state of the at for speech to text for natural language processing. This is like the one part of machine learning I actually care about and you're telling me OpenAI has one of the best most portable models? Extremely discouraging stuff 😔
I'm gonna ask this question with as much humility as I can but can someone help me understand the point of pairing #HomeAssistant with an LLM? I've been reading about the voice pipeline and it keeps mentioning LLMs but I don't understand *why*. Are they just the best form of intent engines that we have now? Are they the current best in class solution for natural language processing? I don't want an LLM but I want a capable voice pipeline
A lot good replies here. The best thing I can say is find an app like @surf that lets you subscribe to a custom feed of sources YOU trust or lets you build that yourself. I've got one for myself that includes a bunch of different news orgs present in the responses.
On one hand it's annoying that my daughters iPad is so old that Netflix doesn't seem to be supported anymore. On the other it does reduce my anxiety level of giving her access to it before school. (It's set up to play learning games with limited other access by default). Not looking forward to having to replace it
This stupidity with California's Age Verification law reminds me that my dream job would be something where I can combine working with Free Software/Open Source with public policy buuuut I have no clue how to get there or if that job even exists 😅
Are people really just sharing a screenshot of some rando asking Claude why it targeted a school for bombing and running with their priors?
I have seen the posts suggesting that the United States war planners used AI to find targets for its bombing campaigns and it erroneously named a school as a legitimate target but I have only seen this be repeated by the same AI that is supposed to have been used in the first place. Has anyone actually seen credible reporting on this?
I love @fontawesome commitment to the bits in their Kickstarter videos. Nice hotdog reference in yet another one. Congrats on the rebrand @11ty 👏
"The pattern is eerily reminiscent of how the xz-utils supply chain attack began, and the security community is still on high alert after that incident nearly succeeded." 😬😬😬
AI is going to end very badly for us.
There's a post I've seen a few times now about how LLMs are a new form of commons enclosure and that if only we had a few sources that weren't controlled by wannabe AI empires they would be this huge democratizing force. And I can't help think of a chapter I just read in Blood in The Machine where the one Luddite leader has the exact same debate 200 years ago with another would be reformer who wants to reorganize society around the machine and build factory towns.
I'm hoping that this is something that folks here in the Fediverse can help me with. My wife is a research scientist in #neuroscience and looking to add some sort of #MachineLearning experience this year to diversify her skill set and maybe get a new job. Since I'm a software engineer she is hoping I can help. As someone who works primarily on web software I have no clue where to get started on this...
I would absolutely appreciate advice and boosts 🙏
I am grateful to Heidi and others who have called out this bullshit but I think it is extremely important that other men also participate in calling out this attitude. We need to hold each other accountable and it cannot always be other women setting the cultural expectation.
@heidilifeldman https://mastodon.social/@heidilifeldman/115645893093480348