@arestelle@dice.camp
@arestelle@dice.camp avatar

arestelle

@arestelle@dice.camp

Lover of language, knowledge nut, shutter snapper, musicophile, stargazer, bookworm. Misfit programmer by day. | she/her | #NoAI

This profile is from a federated server and may be incomplete. Browse more on the original instance.

vasilis, to random
@vasilis@social.vasilis.nl avatar

Dear native English speakers, when do, and when don't you use the word "minutes" when you tell the time?

For instance, would you say “It’s two to two” or would that always be “It’s two minutes to two?”

Wondering when I can leave out the minutes, and when I should definitely use them. Any ideas?

arestelle,
@arestelle@dice.camp avatar

@sarajw @vasilis interesting examples! I wouldn't have thought of those examples where I would say minutes, I think because I would probably say the time clock style when the minutes are not a multiple of 5 - two-twelve vs twelve minutes past two

sarajw, to random
@sarajw@front-end.social avatar

Natalya Shelburne just burst onto the stage with so much energy but with a completely open heart.

"Channeling Chaos: Role of the Artist in the Age of AI"

She's going to convince us that AI does not mean the end of art. She's an artist. Let's go!

arestelle,
@arestelle@dice.camp avatar

@sarajw I'm pretty convinced the time scarcity mentality is itself a problem. Throwing supposed time savers at it is only going to make it spiral faster. We need to step the hell back.

arestelle,
@arestelle@dice.camp avatar

@sarajw yeah, like if you're familiar with Deven Rue (fantasy cartographer) she's talked more than once on her podcast about how she could find AI super useful if there were a way to use it ethically (which there currently isn't - no consent or compensation or way to credit artists exploited to build it - so she doesn't). She's blind in one eye and has no depth perception so it could be really useful to her to generate reference images to work with for things like perspective and shading.

KathyReid, to stackoverflow
@KathyReid@aus.social avatar

I just issued a data deletion request to #StackOverflow to erase all of the associations between my name and the questions, answers and comments I have on the platform.

One of the key ways in which #RAG works to supplement #LLMs is based on proven associations. Higher ranked Stack Overflow members' answers will carry more weight in any #LLM that is produced.

By asking for my name to be disassociated from the textual data, it removes a semantic relationship that is helpful for determining which tokens of text to use in an #LLM.

If you sell out your user base without consultation, expect a backlash.

arestelle,
@arestelle@dice.camp avatar

@kellogh @KathyReid I wouldn't count on that, but they could also remove PII (anonymize username etc) while keeping info about the user's ranking, or likely even about the same user having made all your posts, just anonymize/remove the fact that that user was you - just replace your name with anonymous_whatever

(based on my knowledge, which, I did work on a data platform team at a tech company working toward gdpr compliance with European teammates, but they're the more knowledgeable ones)

emilymbender, to random
@emilymbender@dair-community.social avatar

This is TODAY!

Joins us Monday for the next live stream of Mystery AI Hype Theater 3000:

Annalee Newitz and Charlie Jane Anders of @ouropinions will join me and @alex for a take-down of attempts at principles to govern the use of autonomous agents in Asimov's science fiction and Biden's US military.

Mon, Feb 19, noon Pacific:
https://www.twitch.tv/dair_institute

arestelle,
@arestelle@dice.camp avatar

@emilymbender @alex I'm just listening to the latest YANSS podcast (after which I will now be unsubscribing) and I feel like it IS AI hell - it's all about chatgpt, and one of the guests tells an anecdote about how his grandmother asked what chatgpt is and mentioned that she's thinking about whether she should enter an assisted living facility & by way of answering her, he had her ask it about that and urged her basically to anthropomorphize the hell out of it

emilymbender, to random
@emilymbender@dair-community.social avatar

Next up on Mystery AI Hype Theater 3000: @alex and I will take on the hype around AI and productivity: GPTs are not GPTs; "AI" is not here for your job ... but it might make it shittier.

Join us live at noon Pacific today (Monday Jan 22):
https://www.twitch.tv/dair_institute

arestelle,
@arestelle@dice.camp avatar

@emilymbender @alex the crimes against graph design in this one, my gods

I'm almost surprised they didn't manage to cram some pie charts into that one graph, too

edit: aaand Duolingo just got itself uninstalled. Gross.

emilymbender, to random
@emilymbender@dair-community.social avatar

The tech bros are aware of and even bragging about automation bias:

“Sometimes when our users got used to our product, they would start just kind of blindly trusting it,” said Eli Ben-Joseph, co-founder and CEO of Regard, a company that boasts 1.7 million diagnoses made with its tech, which embeds into a health systems’ medical records.

https://www.politico.com/news/2023/10/28/ai-doctors-healthcare-regulation-00124051

arestelle,
@arestelle@dice.camp avatar

@emilymbender "might yield benefits" and might kill people. How do they get this with drugs but not delegating decisions to machines?

arestelle,
@arestelle@dice.camp avatar

@emilymbender it's especially mind-boggling to me that the industry that had Watson, actually designed to try to get answers right, ages ago is all "but chatbots are the future!" now

timnitGebru, to random
@timnitGebru@dair-community.social avatar

“Andreessen Horowitz is warning that billions of dollars in AI investments could be worth a lot less if companies developing the technology are forced to pay for the copyrighted data that makes it work.” Via neil turkewitz.

https://www-businessinsider-com.cdn.ampproject.org/c/s/www.businessinsider.com/marc-andreessen-horowitz-ai-copyright-2023-11?amp

arestelle,
@arestelle@dice.camp avatar

@timnitGebru "Imposing the cost of actual or potential copyright liability on the creators of AI models will either kill or significantly hamper their development."

As it goddamn should!

timnitGebru, to random
@timnitGebru@dair-community.social avatar

The grift the keeps on giving of the TESCREAL bundle when Anthropic was announced with its founders who left OpenAI because it wasn't effective altruist enough after their deal with Microfost, and now they announce $4B from Amazon 🙄

arestelle, (edited )
@arestelle@dice.camp avatar

@timnitGebru when did EA get taken over by silicon valley billionaires/AI nonsense (or has it always been)? this is so weird to me, having learned about it in a Singer MOOC like a decade ago & mostly knew about Giving What We Can & GiveWell & The Life You Can Save, so like, tithing but not tied to religion and giving that tithe to charities that have evidence showing their work improves lives (now) relatively cost-effectively, which is...the opposite of this?

arestelle,
@arestelle@dice.camp avatar

@timnitGebru ...I just peeked down a rabbit hole about MacAskill and yikes

timnitGebru, to random
@timnitGebru@dair-community.social avatar

So the "AI Safety" people at the chatbot company will not be the ones who think about, I don't know, the ELIZA chatbot from 60 years ago and the issues that arose let alone with this one?

arestelle,
@arestelle@dice.camp avatar

@timnitGebru someone in a discord I'm in recently shared the output they got from what is clearly a wrapper on an LLM (a la ELI5 but more options; "explain like X") where they prompted it for "the most intuitive" description of BPD and felt like its output was super validating and other people chimed in agreeing that's it's fair/realistic/accurate/affirming. I'm not sure if they realized it's an LLM (surely they must? but?), and I don't want to yuck their yum, but 😓

timnitGebru, to random
@timnitGebru@dair-community.social avatar

I just realized that "chief AGI scientist" is a thing at Deepmind when I saw the TIME 100 AI list where Shane Legg who is deep in the TESCREAL bundle, was listed.

arestelle,
@arestelle@dice.camp avatar

@timnitGebru ...sigh.

timnitGebru, to random
@timnitGebru@dair-community.social avatar

The idea of going to Bluesky really doesn't appeal to me. I heard about how Twitter was functional in spite of him. Then he handed it to Elmo saying he is "the singular person I trust to extend the light of consciousness." 🙄 Not to mention the leaked texts between them where Jack Dorsey was telling him he knows "how much you care" 🤮

arestelle,
@arestelle@dice.camp avatar

@albertcardona @timnitGebru it feels like back when gmail was new - the invite code thing is just a successful marketing ploy. (plus a bit of "everybody's joining everything till we see how things shake out"; but bsky doesn't even have hashtags or DMs yet)

luckytran, to random
@luckytran@med-mastodon.com avatar

The US has messed up its booster rollout. People are cancelling vaccine appointments or are wrongfully being charged because of insurance issues. This is unacceptable. Vaccines should be free regardless of your insurance status. https://www.timesunion.com/health/article/insurance-issues-lead-cvs-others-cancel-covid-18368441.php?utm_campaign=CMS%20Sharing%20Tools%20(Premium)&utm_source=t.co&utm_medium=referral

arestelle,
@arestelle@dice.camp avatar

@luckytran Biden: We have the tools!

Me, past 8 months: So I can get another booster since the last one has already waned a lot?

American "health care": No.

Me: So can I expect people to wear masks so they don't cough COVID in my face when I'm buying groceries?

'murika: hahahahaha no

Me: ...in health care settings, though?

'murika: lol suck it

Me: So, the tools exist in principle and y'all're just gonna pretend they don't have to actually be available in practice. Cool cool cool (fml)

arestelle, to Facebook
@arestelle@dice.camp avatar

The form for deleting your data from FB's AI that we passed around a little while back is not an opt-out in any way, shape, or form.

Turns out it's not even a "delete my data from your training set" form - they will only do anything if you can prove their AI spit out personally identifying information in its output - like your name or contact info.

Never mind everything else, including copyrighted work, & never mind them feeding that data into their models.

arestelle, to fountainpens
@arestelle@dice.camp avatar

Birmingham Pen Company has inks specifically meant for mixing your own colors (or theirs, with recipes), and in celebration of lining up a new job I...only bought five (and the dilution stuff, which was on sale and will let me play with improving the too-dark samples I already had). This merits applause.

I might need to go play, now.

badastro, to random
@badastro@mastodon.social avatar

I am so, so, SO done with Twitter.

https://badastronomy.substack.com/p/goodbye-twitter

arestelle,
@arestelle@dice.camp avatar

@badastro so say we all (and that is absolutely the way I read "xitter" as well, only correct option imo 😈)

timnitGebru, to random
@timnitGebru@dair-community.social avatar

SMH. These manchildren technocrats running the world is just unreal.

arestelle,
@arestelle@dice.camp avatar

@timnitGebru a) 🤦
b) how much is it genuine dumbfuckery, and how much is it egging on the implosion of the competition, I have to wonder

timnitGebru, to random
@timnitGebru@dair-community.social avatar

Meetings should be 30 minutes by default unless there are circumstances that warrant more time. A lot can be done in 30 minutes. That's one thing I agree with Google people on, but then again, people added so many random meetings on calendars without even asking first.

arestelle,
@arestelle@dice.camp avatar

@timnitGebru I'm a fan of the 25-minute meeting, if the end time is actually honored. You can probably get the same amount done, and then you have a buffer to get another cup of tea before the next whatever.

openculture, to random
@openculture@toot.community avatar

A New Course Teaches You How to Tap the Powers of ChatGPT and Put It to Work for You

https://www.openculture.com/2023/08/a-new-course-teaches-you-how-to-tap-the-powers-of-chatgpt-and-put-it-to-work-for-you.html

arestelle,
@arestelle@dice.camp avatar

@openculture No.

Popehat, to random

My post about the latest Trump indictment and about how people -- specifically the National Review -- are lying to you about it.

https://popehat.substack.com/p/people-are-lying-to-you-about-the?sd=pf

arestelle,
@arestelle@dice.camp avatar

@Popehat "Lies are not opinions" 👏

Adrenochrome, to italy German
@Adrenochrome@mastodon.social avatar

A huge cluster of grapes in
By

arestelle,
@arestelle@dice.camp avatar

@Adrenochrome @hoaxeye

arestelle,
@arestelle@dice.camp avatar

@hoaxeye @Adrenochrome I found a reddit post with what looks like the same image; a commenter said that it was a constructed display, so not fake but misleadingly described/hashtagged

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • JUstTest
  • osvaldo12
  • InstantRegret
  • thenastyranch
  • magazineikmin
  • everett
  • rosin
  • Youngstown
  • slotface
  • cubers
  • cisconetworking
  • kavyap
  • GTA5RPClips
  • DreamBathrooms
  • megavids
  • tacticalgear
  • khanakhh
  • mdbf
  • Durango
  • ngwrru68w68
  • tester
  • normalnudes
  • ethstaker
  • provamag3
  • modclub
  • Leos
  • anitta
  • lostlight
  • All magazines
    •