RE: https://flipboard.com/@kotaku/kotaku-latest-2mbnjtrcz/-/a-OsKQVR-yRA29T_t3AYF_ZA%3Aa%3A1938881949-%2F0

software will keep getting worse until people stop using it for important things.

the journalists here aren't exactly covering themselves in glory either. like, not a single one of them knows the law on this???

wow Andy Harris is really crashing out, huh. https://www.reddit.com/r/law/comments/1x60cgz/republican_rep_andy_harris_cites_twoparty_state/

sorry, I am in the middle of reading a bunch of John le Carré novels so this **extra** bothers me

RE: https://mastodon.social/@arstechnica/117440258476982342

this guy sounds like the worst spy ever. seems obvious he was working for someone else, probably under threat. if the FBI were competent, they would have turned him and tried to roll up the whole network. really fucking stupid indictment.

so cool that I have to get this kind of analsys from WallStreetBets https://www.reddit.com/r/wallstreetbets/comments/1x5415v/broad_market_is_breaking_down_while_top_10_big/

Edward Norton is cooler than you and wants to fuck your wife, but only with your permission and in the most respectful way possible.

I've seen the Spanish version and this casting/poster is 😘👌

(hey pst, if Anthropic were profitable, they would have already gone public.)

RE: https://mastodon.gal/@elrohir/117438355557524273

100%. LLMs can't do judgment, just like any API call cannot do judgment. it does a **facsimile** of judgment, which is apparently good enough to fool most C-suit executives. which, hey, really makes you think!!

oh my fucking god I DO NOT GIVE A SHIT ABOUT KELLY CLARKSON WHAT THE FUCK, jesus.

@ebooksyearn in RE: our earlier conversation about blond ladies and extra-dark eyebrows...

it's also why models that are good at math or coding or whatever are worse at other things. machine learning models have **tradeoffs**. you can do reinforcement training to emphasize a particular feature, but that often means sacrificing performance on other features.

either you have a "safe" customer service chatbot that's been so lobotomized you might as well just link people to the FAQ; or you have an engaging, human-like chatbot that occasionally calls customers the n-word.

there's really no way to square that circle, and it's why, now **years** after the launch of ChatGPT, public-facing chatbots are still basically indistinguishable from previous automated help chats.

this gets to a thing I've been saying for years now: you can absolutely turn the temperature up on an LLM to make the output more "creative" or interesting, but that's also what makes it more likely to coach someone into AI psychosis and suicide.

"in attempting, eventually, to make its freewheeling platform safer, [Character AI] also made it unusable and deeply frustrating for much of its userbase — suggesting that the qualities that allegedly made its platform so dangerous to some young users were the same features that made it compelling." https://futurism.com/artificial-intelligence/google-character-ai-bankruptcy

I've had people tell me "oh you're American, so you must like spicy food" folks this is progress 🥲

there's a trope in the US that white people can't handle spicy food, but the ball has been moving for decades now and your average white American regularly eats things that would cause your average Spaniard to curl up and die.

looking forward to Michael Tracy's inevitable hospitalization.

this is a huge reason why people like/follow Ed Zitron: he's funny. the other people "doing the work" have my deepest sympathies, and they absolutely deserve credit where it's due, but a fact of the human psyche is that being funny is a **very** effective communication strategy.