Wonder how the survey was sent out and whether that affected sampling.

Regardless, with -3-4k responses, that’s disappointing, if not concerning.

I only have a more personal sense for Lemmy. Do you have a source for Lemmy gender diversity?

Anyway, what do you think are the underlying issues? And what would be some suggestions to the community to address them?

sounds like this can be a plot of a new Pixar movie

cuts parts

that’s actually what the underlying method does, as this is extractive summary, hence it mostly cuts and stitches things.

From my naive understanding, this type of method does not use or “understand” context.

The alternative is abstractive summary, which is where LLMs (or even small/medium language models) are good for. But I suspect that would be a controversial choice on lemmy.

I’m also curious. A quick search came up with these. Not sure which one is most reliable/updated

• https://www.comparitech.com/blog/information-security/global-ransomware-attacks/
• https://www.comparitech.com/ransomware-attack-map/
• https://cissm.liquifiedapps.com/
• https://cybermap.kaspersky.com/special/ransomware

Many things are called “AI models” nowadays (unfortunately due to the hype). I wouldn’t dismiss the tools and methodology yet.

That said, the article (or the researchers) did a disservice to the analysis by not including a link to the report (and code) that outlines the methodology and how the distribution of similarities look. I couldn’t find a link in the article and a quick search didn’t turn up anything.

you should try to ask the same question using xAI / Grok if possible. May also ask ChatGPT about Altman as well

Lol ads that can be engineered into DNA, so that they can be passed down for generations.

Based on this reddit comment, that website is not affiliated with the magic-wormhole CLI tool

doesn’t seem to be so comfortable with glasses, esp with a hoodie unfortunately

fascinating. I wonder where the line is between the cold preserving the body and the cold causing hypothermia that could lead to death.

I believe experiments like these should move slower and with more scrutiny. As in more animal testing before moving on to humans, esp. due to the controversies surrounding Neuralink’s last animal experiments.

re: your last point, AFAIK, the TLDR bot is also not AI or LLM; it uses more classical NLP methods for summarization.

Speaking of Lost, its parody show called Wrecked ended too soon.

Is there a database tracking companies that start out with good intentions and then eventually gets bought out or sells out their initial values? I’m wondering what the deciding factors are, and how long it takes for them to turn.

Daredevil (the design and music is sick) and The Morning Show (the animation is very captivating to me)

re 1: out of curiosity, do you encounter dnsleaks when using wireguard?

re 4: you can also check out https://starship.rs/, which helps configure shell prompt very intuitively with a toml file.

Reminds me of this article https://www.alexmurrell.co.uk/articles/the-age-of-average where the author pulls in different examples of designs and aesthetics converging to some “average”.

I’m feeling conflicted with these trends, on one hand it seems like things are becoming more accessible, while on another, feels like a loss.

This especially may be relevant with generative AI - at least for the very few generative arts I look at, at some point they start to feel the same, impersonal.

“Bad” can be quite broad and it might be cumbersome to check and categorize all of the “badness” out there. You might have better luck narrowing down a bit. For example, if you’re interested in AI/algorithm incidences, there are at least two that came up on search:

• https://www.aiaaic.org/aiaaic-repository
• https://incidentdatabase.ai/

On a tangential note of another comment about AI training and such, this is a touchy and evolving subject, but it might be good to include how you want your content to be used and not be used, and by whom, especially if you intend to make them public.

some wiki backends allow password protection. for example, mkdocs, which also renders markdown, has mkdocs-encryptcontent-plugin to allow global or even page-specific passwords for private repos.

but these encrypted pages would of course have the risk of not being archived by the wayback machine.