As former Project Manager and people manager, my reaction to “Simply put, there is no way in which one can rely on an unreliable tool, especially when it robs us of the very ability to recognize when it's wrong.” Is “lol lmao”

it is *so* loud here. luckily, slam is unfazed. He’s sitting in my lap watching the fireworks. (Sound on)

i really appreciate that recent marvel rivals updates have allowed me to use the word 'dickbone' more often.

the ceo of doordash is worth 1.8 billion. the average door dash driver makes less than $15/hour. i suspect people are advocating for better wealth distribution among the working class more so than paying delivery drivers less, but i could be wrong ¯\_(ツ)_/¯

Your dog hates fireworks. My dog is chasing me with a roman candle because she celebrates liberty

so many fireworks are being set off, and it's only 5 pm. ah yes, the smell of freedom: gunpowder.

it's rare that i vibe code, but when i do, i always tell the agent "make good decisions, because you're the one that's going to be stuck maintaining this and adding new features" and i wonder if that actually has any real effect on output.

tbh this is probably more a reflection of neurotypical vs neurodivergent communication styles (just a guess, based on the comments). most of the time, devs aren't saying "you're too stupid," they are just trying to not be misunderstood >.< there are of course exceptions.

i tried this on glm-5.2.

in 2027, some 13-year-old is going to learn about a band called "Nirvana" for the first time thanks to this yuri anime and I think that's beautiful

half the people on my nextdoor are using ai to write their posts. it's actually insane. they don't even edit the formatting. and then they also comment using ai. it's not even claudeisms. it's all chatgpt bullshit. ffs i do not want to talk to chatgpt. i barely want to talk to my neighbors.

this whole thread. my lort. somebody save me

i have been dealing with the fireworks since the 28th of june. i'm not gonna tell people to not set off fireworks for the 4th of july. i live in a tourist beach town. but jfc can you save it for the ACTUAL 4th of july and not be dumbasses the entire week before, i beg.

Skill issue.

I'll summarize this: Mamdani asked NYers to keep air conditioning at 78 degrees. A wave criticism hit online. A MSNBC broadcast pointed out the US Department of Energy recommends that 78 degree figure. Immediately after, the US Department of Energy deleted 6,000 web pages related to energy saving.

my mom is walking around my house with the handyman talking about caulk.

pulling 50 tps from umans but prompts still take 5 minutes because glm-5.2 is a verbose mf

"fable actually tells me no" i mean, opus does this too. gpt also does. glm also does. if you need to spend $50/mtok to get the model to tell you no, you're using it wrong.

i was today years old when i found out that joseph wasn't dale's son in king of the hill. sigh.

one of my old osu friends told me he found a replacement for osu in some game mode in roblox and told me i needed to join him. i am nearly 45 years old. there is no way in hell i'm getting on roblox. that is the one game i will not play. not at my age, ain't no way.

seriously this is some of the worst stuff I have ever seen, and they have done a lot of incredibly bad things github.com/anthropics/c...

what happened to popehat?

why is anthropic making their models progressively more annoying every release? i spent 2 minutes talking to sonnet 5 and have no desire to use that model ever again.

i think i made a mistake i was grooming my golden doodle, giving him his summer cut with a 5 blade, and his tail was super long and i couldn't cut it straight. so i figured, why not shave his tail? I THOUGHT IT WOULD BE OK. it is not ok. it looks like a rat tail. he won't look at me.

Peak bluesky content in a post about synthetic cheese sauce.

there are not enough tank girl gifs

Thought for 17s

idk how it's possible to be mostly e/acc but also evenly split on pro/anti ai but here i am. i think this is actually more an indication of how much i distrust our government and corporate model alignment efforts, which is somehow read as e/acc.

@chriswarcraft.bsky.social bro

thinking about writing about community driven moderation vs ai moderation vs traditional (human, paid) moderation. I used to think community driven moderation with protocols to prevent standards drift was the ideal system. not perfect, but better than alternatives. now, that may not be the case.

its a shame community notes is only on Twitter

mildly concerned. is this gamer food? this seems like gamer food.

power consolidation is rarely good for the people. i'd like to think that dario is looking at this more from a safety than a financial reason, but the end result is the same. open source ai is what keeps me optimistic. if openai/anthropic held the keys, the situation would be dire.

i use gh issues to track work that needs to be done after code-reviews, new features i have planned, etc. my projects tend to be more personal and solo-contributor. i left some issues open overnight, and i came back to a bunch of waiting PRs. while some of the code isn't bad, this is concerning.

people are scraping github looking for issues to work on with models that make indentation errors, that's crazy.

My personal precarity hazard is elevated, more than usual. Ten patrons joining the current five at the present average of $10/month would get me up to where I was able to handle it the last time I was in the situation I'm concerned about. Reposts and likes also help!

WELL YOU OPENED 15 ISSUES AFTER A CODE REVIEW, KIMI. i'm not made of stone.

It was inevitable

"If the Supreme Court is willing to deny key freedoms to some of the nation’s most vulnerable by narrowing its interpretation of whom the Constitution applies to and protects, it stands to reason that it can narrow it further for others,

stop arguing with the guy who loves to argue

most serious projects or production software require prs to be reviewed so that code always sees at least two sets of eyes on it. idk why having an LLM review another LLM's work is something folks think is weird.

One of my hot takes: if the AI economic transformation pans out, a robust humanities education will be a tremendous competitive advantage. The capabilities and risks of LLMs are dramatically different depending on the user's critical thinking, reading, and writing skills.

16 years ago @theonion.com

i generally avoid participating in conversations around bluesky main character of the day but i have so many opinions about this one even though bluesky is not the proper venue for me to express them because they are so complicated. :|

Restrictions on open source are a restriction of speech and a government mandated monopoly (or duopoly), and vulnerable to disruption by those who aren't bound by the same regulations globally. It should be illegal for Mythos to be closed, the weights should be guaranteed and required to be open

i take it back. don’t stop air conditioner discourse. i’d rather us have air conditioner discourse than have “do you have a license for that llm mate?” discourse

There are major governments who think boys kissing creates significant social risk. Or playing DND, or being on birth control or touching your own genitals, or marrying a person a shade darker than you. When you propose authoritarian measures you will always always always get authoritarian results.

glm-5.2 doesn't seem to have the same conservative behaviours that have been RLHFed into newer frontier models when it comes to proactive tool use. the cost of the alignment tax is noticeable when you're given an alternative. very cool to see (at least, for my use case).

my computer is over there, scrolling away on it's own. and i hear, while i'm making dinner, from my computer speakers: "as a senior platform engineer and a bisexual, i refuse to choose between ipv4 and ipv6" i mean happy pride i guess?