David Gian-Cursio

@davidcgc
162 Followers
854 Following
806 Posts

3D guy. Mac guy. Sci-Fi guy. Film guy. He’s just this guy, you know? [He/Him]

Avatar by Dan Jones

Websitehttps://www.gian-cursio.net
Apparently, the Artemis II has 260 Mbps internet via lasers, I guess technology has improved a little bit since the last time we were visiting the moon.
Believe it or not, there's already a mechanism in place for the federal government to seize and appropriate the assets of billionaires and redistribute them for the public good. SCOTUS accidentally created it with one of its worst decisions of the 21st century, Kelo v. New London.
The thing I love about the stock market is that when it does well, I don’t get anything, but if it does bad, everything pays less and costs more.

Observation by @jonny of some of the insanity in Claude's source code:

"Apparently it is difficult to make this actually happen though, as the parent LLM likes to launch the forked agent and just hallucinate a response as if the forked agent had already completed."

This is a perfect refutation of both the argument that they are somehow doing more than predicting the most plausible next word AND of the argument that it doesn't matter if that's all they are doing.

https://neuromatch.social/@jonny/116331911643580333

jonny (good kind) (@[email protected])

Attached: 1 image I seriously need to work on my actual job today but i am giving myself 15 minutes to peek at the agent tool prompts as a treat. "regulations are written in blood" seems like too dramatic of a way to phrase it, but these system prompts are very revealing about the intrinsically busted nature of using these tools for anything deterministic (read: anything you actually want to happen). Each guard in the prompt presumably refers to something that has happened before, but also, since the prompts actually don't *work* to prevent the thing they are describing, they are also documentation of bugs that are almost certain to happen again. Many of the prompt guards form pairs with attempted code mitigations (or, they would be pairs if the code was written with any amount of sense, it's really like... polycules...), so they are useful to guide what kind of fucked up shit you should be looking for. so this is part of the prompt for the "agent tool" that launches forked agents (that receive the parent context, "subagents" don't). The purpose of the forked agent is to do some additional tool calls and get some summary for a small subproblem within the main context. Apparently it is difficult to make this actually happen though, as the parent LLM likes to launch the forked agent and just hallucinate a response as if the forked agent had already completed.

neurospace.live
I just lost an hour to this thread. I miss old twitter.

RE: https://bsky.app/profile/did:plc:bii2o576sszvunaymlwnjfmu/post/3mihzso3gcc2j

RE: https://mastodon.social/@zadr/116304654036821564

I wonder if this could be done for “Jelly” and the other two then-new iTunes visualizers that didn’t survive the transition to 64-bit.

Since there seems to be #Firefly in the air, here are some renders of I made of Sean Kennedy's #Serenity model back in the day.

I've got a copy of his tutorial to model the film version, still. One of these days…

If you've been having trouble finding a tech job in the US over the past few years, here's some data to validate that it isn't just you.
Hi. Please play this amazing JND color game my friend Keith made and see how good you/your displays are at differentiating between similar colors. https://www.keithcirkel.co.uk/whats-my-jnd/?r=filmgirl
What's My JND?

Find your Just Noticeable Difference in colour perception. How small a colour difference can you actually see?

Almost missed my semi-annual complaint about #AppleWatch’s Solar Dial being unable to cope with daylight savings time changes and losing alignment on the 23/25 hour days.