social.bund.de is one of the many independent Mastodon servers you can use to participate in the fediverse.
Dies ist der Mastodon-Server der Bundesbeauftragten für den Datenschutz und die Informationsfreiheit (BfDI).

Administered by:

Server stats:

100
active users

#artificialintelligence

199 posts96 participants2 posts today

Media — March 2025

Reading

The Giver Graphic Novel

I haven’t read the novel since middle school but I remembered really liking it a couple of times. Kid got this from the library so I read it quickly… there was a lot I had forgotten beyond the basic world building concept. Fun to refresh my recollection in a different medium.

Writing After the Homework Apocalypse

This piece on writing to demonstrate learning becomes less useful in a world with widespread generative AI, but that writing itself as a process still has significant cultural value.

Modern cultural labor is a process of responding to and pushing against the values represented in existing culture, but it is also a process of adapting to structural and technological change.

Scaling People

This was an interesting work read recommended to me by someone in our book club who also works in cybersecurity. It’s definitely geared more for a director-level audience but I found a few things that were useful. (I also skimmed and skipped a lot of the sections on hiring and promotions, since those processes are defined company wide and outside of my influence.)

A few things that jumped out at me:

Success depends on all team functions coming together to deliver a great product

p. 84

The worst thing you can do is let a temporary structure persist for too long.

p. 264

Which I interpret together as calls to create the right kinds of cross-functional teams for a project, which includes way more than just software engineers.

And additionally since I live the hybrid meeting lifestyle working with many team members not co-located with me in Seattle:

Give remote work equal footing in your organization.

p. 313

Playing

Fallen Order

I’d heard a lot about this game and where it fits into some of the stuff we saw in the Obi-Wan series. I haven’t played a Star Wars game outside of the Lego ones in some time.

Kid hasn’t seen a lot of AAA games like this (being largely a Switch/iOS household) and even as old as it is he was really impressed by the realism of the graphics. I guess we’ve both been missing out!

I’m pretty bad at some of the 3D jump quick time events and blocking, but it’s still fun and an interesting Star Wars story so far. Sure seems like a lot of Jedi managed to survive Order 66

Grindstone

With my wife and kid playing again, I went back to one of my favorite Apple Arcade games. I apparently haven’t touched it in over three years! I’m trying out the new Carnival of Creeps set of levels.

Watching

We just finished Severance S2 in early April so I’ll include that in next month’s post. Still almost too stressful to watch before bed!

Across the Spider-Verse

I forgot this was a Part One until about 15 minutes before the end when I realized it wasn’t wrapping up. Family was a little upset about that. Still, an amazing movie, even if it doesn’t quite stand alone.

Really it’s a parenting story that happens to also be a superhero story, coupled with some absolutely astounding visuals. There’s just so much coming of age story and relationship story wrapped up in the meta commentary about Canon Events and the traditional superhero tragic backstory. I especially like how Miles says “no” to having to accept his fate just because it’s The Story.

Visual wise I particularly loved the underground zine style of Spider-Punk and the da Vinci drawing style of Renaissance Vulture, but everything was so well done from character designs to backgrounds and more. I’m still thinking about it. I’m surprised it didn’t win more awards but it was also a stacked year for the Academy (Elemental, Nimona, and the winner The Boy and the Heron).

I have to imagine the alternate world where the MCU was animated like this instead of being live action. Maybe they wouldn’t have been as big of money makers, but at least you’d be able to actually see all the action and lean into the best elements of the original comic book format.

Very excited for Beyond, even though it just got a release date… in 2027!

Listening

Lady Gaga — Mayhem

This album is amazingly good. We’ve listened to it through easily a dozen times since release, and it’s a lovely sendup of dance pop over the last few decades. More knowledgeable people than me have analyzed the influences but I can hear both callbacks to her own oeuvre as well as honoring influences like Prince, Michael Jackson, and many more. I heard some Savage Garden in there and so did Darren Hayes. We are of course particularly primed to like the Swift-sounding “How Bad Do U Want Me” that could pass for a 1989 vault track, but I have had several others stuck in my head including “Killah”, “Don’t Call Tonight”, and of course “Abracadabra”, which I highlighted last month.

Lingthusiasm — The Science and Fiction of Sapir-Whorf

A nice overview podcast episode of both functional stories where it plays a role and the linguistic research that has largely weakened the hypothesis. It’s an idea that always sounded really cool when I was first getting interested in language.

The Llama 4 herd: The beginning of a new era of natively multimodal AI innovation.

programming.dev/post/28113022

programming.devThe Llama 4 herd: The beginning of a new era of natively multimodal AI innovation. - programming.dev> - We’re sharing the first models in the Llama 4 herd, which will enable people to build more personalized multimodal experiences. > - Llama 4 Scout, a 17 billion active parameter model with 16 experts, is the best multimodal model in the world in its class and is more powerful than all previous generation Llama models, while fitting in a single H100 GPU. Additionally, Llama 4 Scout offers an industry-leading context window of 10M and delivers better results than Gemma 3, Gemini 2.0 Flash-Lite, and Mistral 3.1 across a broad range of widely reported benchmarks. > - Llama 4 Maverick, a 17 billion active parameter model with 128 experts, is the best multimodal model in its class, beating GPT-4o and Gemini 2.0 Flash across a broad range of widely reported benchmarks, while achieving comparable results to the new DeepSeek v3 on reasoning and coding—at less than half the active parameters. Llama 4 Maverick offers a best-in-class performance to cost ratio with an experimental chat version scoring ELO of 1417 on LMArena. > - These models are our best yet thanks to distillation from Llama 4 Behemoth, a 288 billion active parameter model with 16 experts that is our most powerful yet and among the world’s smartest LLMs. Llama 4 Behemoth outperforms GPT-4.5, Claude Sonnet 3.7, and Gemini 2.0 Pro on several STEM benchmarks. Llama 4 Behemoth is still training, and we’re excited to share more details about it even while it’s still in flight. > - Download the Llama 4 Scout and Llama 4 Maverick models today on llama.com [http://llama.com] and Hugging Face. Try Meta AI built with Llama 4 in WhatsApp, Messenger, Instagram Direct, and on the web.