One fun element is the double lift card trick reveal.

Like get their card ontop somehow and the descretly lift the top 2 cards up, pretending they are just a single card. Then you show them the 2nd to top card and ask is this your card? They will say no, you put the two cards on the top of the deck, do a magic tap or something, and then you only lift the top card, their card off the top of the deck after it transforms into theirs

I think I used to do a similar one, but with memorizing the top card of the deck and having them cut the deck! Lots of variations too and presentation is everything

I don’t know why isn’t it just the parents responsibility. For example where I live, you can get charged with “negligent storage of a firearm”. Why can’t there just be a “negligent allowing your kid to do whatever they want with no DNS filtering” offense

Not until you hit your token limit ;)

Not necessarily. If it answers you question and go go about your day, that’s not as good for Microsoft user numbers/engagement.

They probably want it to just barely get you exactly what you need to keep you on the edge. Once they start training models for user engagement the enshitification will begin

They just want it to he addictive so that people use it a bunch. Then they try to sell enterprise level versions to companies they can say look how much people use it, it must be because it is so useful

Cool concept in theory, SUPER annoying and risky because in federal use because there is no “undo button” for all the mistakes the agent will make though

For really lightweight models, Qwen3:8b is pretty good for an 8gb graphics card.

Gemma 4:e4b is pretty good too. I usually sun that on my 16gb gpu.

Obviously the little ones aren’t as good as big ones, but you can always rely on real intelligence to fill in the gaps

$2 per year for my domain name.

I actually support this as long as the false positives are not too bad

Privacy is important because laws and “normal” can change. What might be normal and legal today, could become illegal someday. Then suddenly all your data that used to describe you as a boring normal person, now describes you as a “domestic terrorist”

I’m still learning FreeCAD, but so far is seems just fine to me! Just different from what I am used to. But there are a few good channels on YouTube that do good tutorials

Yeah I was curious about that. Like I can poke around in the immich directory, but the actual pictures are stored in a weird structure. Do you have any recommendations?

One of my first self hosting projects was a jellyfin server. Double check, but I think the main hardware requirements are just 4GB of RAM and enough harddrive space for your videos/files!

I really like immich too. It’s like Google photos, but self hosted. It’s super fast for uploading and backing up your photos over your local network. Immich also needs at least 4GB of RAM I think

I’m with ya there. I think a lot of the valid hate towards AI is is actually for Big Tech companies and data centers.

I support folks running less powerful locally hosted models on their own hardware, which I also do myself!

60 is such a nice highly composite number though!

https://en.wikipedia.org/wiki/Highly_composite_number

Ah yeah, I think you need a 6GB gpu for images and probably like a 12Gb GPU for videos.

I think you can run with models without, but it just slows way down

It’s kinda fun to poke around with image and video generation using ComfyUI. There’s all sorts of models, LORAs, and setting to tune and play with

Noob here, what’s wrong with curl | sh installs?

Grok is dumb because it is one of the least efficient models out there. If you’re just gonna do pointless role-playing use a smaller model that is fine tuned for role playing