@dweinberger@mastodon.social
@dweinberger@mastodon.social avatar

dweinberger

@dweinberger@mastodon.social

I write about the way digital tech - especially #AI these days - may be changing our ideas about ourselves and our world. #GrandKids #Philosophy #MachineLearning #Internet #AI #LeftyBoomer #Ethics #EthicsOfCare #Cluetrain #BKC #Internet #Web #peace. https://www.weinberger.org/writings/index.html 35% yogurt by volume.

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

dweinberger, to random
@dweinberger@mastodon.social avatar

Robert Dennard, who invented DRAM, has passed away.

May his memory be a blessing.

https://www.nytimes.com/2024/05/16/technology/robert-dennard-dead.html

dweinberger, to ai
@dweinberger@mastodon.social avatar

Here is the link to the article @garymarcus
discusses that argues that the relationship between training set size and improvements in models' performance is non-linear in the wrong direction. https://arxiv.org/abs/2404.04125

dweinberger, to random
@dweinberger@mastodon.social avatar

Philosophies has published my article "The Rise of Particulars: AI and the Ethics of Care." https://www.mdpi.com/2409-9287/9/1/26

In brief: ML discovers patterns of correlations, a powerful form of generalization very different from the sort that the west has valorized as the highest form of truth: universal laws, ethical frameworks, etc. ML's generalizations let particulars speak loudly, creating an odd analogy with the feminist ethics of care.

dweinberger, to random
@dweinberger@mastodon.social avatar

Cerebras Systems announces world's fastest chip. 2,048 can be clustered into a supercomputer capable of 256 exaFLOPS = 256,000,000,000,000,000,000 = a quintillion Floating Point Ops per second.

Should hold me for now.

https://newatlas.com/computers/cerebras-wse3-cs3-worlds-fastest-ai-chip/

dweinberger, to ai
@dweinberger@mastodon.social avatar

Research showing LLMs' (chat AI) covert racism, triggered by dialects, is even worse than their explicit racism. An example of when you can be shocked but not surprised. via Gary Marcus https://garymarcus.substack.com/p/covert-racism-in-llms

dweinberger, to ai
@dweinberger@mastodon.social avatar

After a long-ish exchange, chatGPT 4.0 acknowledged that it speaks from a Western, Northern cultural position. Obviously. Not as obvious is how “reluctant” it is to acknowledge that fact.

And having acknowledged it, it will deny it the next time. "I acknowledge x" is just a set of tokens chosen to get us to move on, like a politician's apology.

https://www.hyperorg.com/blogger/2024/03/03/chatgpt-admits-it-assumes-a-western-northern-culture/

dweinberger, to random
@dweinberger@mastodon.social avatar

Does anyone know where the unsub button is on the Patreon web site? I want to unsub from some projects I've been minorly supporting. I can't find a button anywhere, or any support or help options from Patreon. So now I not only feel stingy, I feel dumb.

dweinberger,
@dweinberger@mastodon.social avatar

@thadd @shoq OMG that is so hard to find, misplaced, tiny, and shady. Thanks so much, Thadd!

dweinberger,
@dweinberger@mastodon.social avatar

@shoq @thadd Great project! But in fairness, Patreon does give a more obvious way which I somehow missed. Still, the way @thadd pointed out is obscure and misleading: a pencil means "edit". The only edit this particular pencil enables is the one a trash can usually indicates.

dweinberger, to random
@dweinberger@mastodon.social avatar

I received an invitation from ISONAET to speak at its Summit on Applied Science, Engineering & Technology at a cost of a mere $800. The invite email explains: "We were impressed with your work and the way you are so impressed at that thing you do and would be so excited if you would join us."

I don't mean to brag, but who wouldn't be impressed by that thing that I do!

Seriously, has anyone heard of this group and its overwhelmingly male roster of speakers? https://scitechglobalmeets.com/2024/applied-science/

dweinberger, to ChatGPT
@dweinberger@mastodon.social avatar

4 and do terribly when asked to find words that have a consecutive M & L. They got “gimlet”, probably because it was in my prompt. Then they got 10-12 words that have M and L's in them, but not adjacent. I guess that's what happens when every word looks like a token

dweinberger, to ai
@dweinberger@mastodon.social avatar

"The 5 Ages of Modern ", -- from reduced and normalized resource to the holder of secrets and mysteries-- is the topic of my latest column in @KMworld, based on my keynote at the KMWorld conf. https://www.kmworld.com/Articles/Columns/Perspective-on-Knowledge/The-five-ages-of-data-161937.aspx

dweinberger, to random
@dweinberger@mastodon.social avatar

Today I learned: Hydox was the original and Oreo is the knockoff! Why isn't this bigger news? Just goes to show the power of Big Cookie. https://thehustle.co/the-curious-case-of-the-disappearing-hydrox-cookies via @mathewi @journa.host's excellent daily aggregator "When the Going Gets Weird" https://newsletter.mathewingram.com/

pluralistic, to random
@pluralistic@mamot.fr avatar

I'm kickstarting the audiobook for The Bezzle, the sequel to last year's Red Team Blues, featuring Marty Hench, a hard-charging, two-fisted forensic accountant who spent 40 years in Silicon Valley, busting every finance scam hatched by tech bros' feverish imaginations:

http://thebezzle.org

--

If you'd like an essay-formatted version of this thread to read or share, here's a link to it on pluralistic.net, my surveillance-free, ad-free, tracker-free blog:

https://pluralistic.net/2024/01/10/the-bezzle/#marty-hench

1/

dweinberger,
@dweinberger@mastodon.social avatar

@pluralistic You had me at "hard-charging, two-fisted forensic accountant." @pluralistic never disappoints.

dweinberger, to random
@dweinberger@mastodon.social avatar

You want AI Weirdness? @Janellecshane has AI Weirdness for you! The latest: ChatGPT generates labelled images of mammals, in English, Swedish with too many umlauts, and English with too many umlauts. (Her book You Look Like a Thing and I Love You is an excellent + funny ML 'splainer.)
https://www.aiweirdness.com/learn-the-mammals-with-dall-e3/?ref=ai-weirdness-newsletter

dweinberger, to random
@dweinberger@mastodon.social avatar

[Edit: Aaaggh. I was misled by my lying eyes. Also, I'm a dolt. How embarrassing!]

Twitter is requiring me to reauthenticate, which requires solving 5 easy but annoying addition problems in which you have to pick the dartboard (out of ten) that has numbers that add up to the target number. Except not all of them have solutions, which requires starting all over again. So far, I haven't found one in which all 5 challenges are answerable. (Why am I still on twitter? For reasons.)

Screen capture of Twitter presenting 10addition problems that don't solve the the challenge.

dweinberger, to journalism
@dweinberger@mastodon.social avatar

@jeffjarvis' heartfelt, impassioned call for to move from seeing itself primarily as an information service, to being a builder and enabler of empathetic, truth-based communities. (Disclosure: We're friends. He doesn't know I'm posting this.) https://medium.com/whither-news/journalism-belief-belonging-07a8dabe35e4

dweinberger, to ChatGPT
@dweinberger@mastodon.social avatar

Chomsky says Chat AI (e.g., ) copies "existing works from existing artists" and alters "it sufficiently to escape copyright laws." That's no more true than saying that someone who records which birds are seen together is copying existing birds.

dweinberger, to random
@dweinberger@mastodon.social avatar

John Searle wrote "...I am told that some of the ancient Greeks thought the brain functions like a catapult." I spent some time this morning doing shitty Internet research and can't find anything, except a bulleted list of lies from Bard. (ChatGPT couldn't be bothered to make anything up, to its credit.) Any suggestions? (I do not read Greek, old or new.) Thanks!

dweinberger, to random
@dweinberger@mastodon.social avatar

@stevenlevy, one of my favorite writers about tech, has posted a good old-fashioned “rant 'n' take down” of Andreessen's perfect example of what it means to be blinded by privilege.

Andreessen: https://link.wihttps://a16z.com/the-techno-optimist-manifesto/

Levy: https://link.wired.com/view/5be9d5af2ddf9c72dc1f8e2bjp4yb.nx/e390c36a

dweinberger, to random
@dweinberger@mastodon.social avatar

After using MS word for almost 40 years, it remains buggy in its basics I won't bore you with a list (you know who you are, AutoNumbering!) but tonight its refusal to let me join or separate two lines broken in the middle is particularly annoying. You'd think by now Word would have mastered line and paragraph breaks, which are pretty basic document elements. I mean, it's not like they're as wildly complex as SIMPLE STUPID AUTONUMBERING A SIMPLE STUPID LIST.

dweinberger,
@dweinberger@mastodon.social avatar

@JamesGleick I feel like I shouldn't be laughing, but it's such a painful, and distressingly relevant piece. Thanks.

I'd been using bug-free Wordstar for 5 years or so when I went to work in 1986 for a fully wysiwyg, typeset-quality, text and graphics wordprocessing company — Interleaf — that successfully handled far more complex , structured layouts than Word now. Yes, it had bugs, but it fixed them. And autonumbering worked.

These days I'm back to WordStar's equivalent: markdown.

dweinberger,
@dweinberger@mastodon.social avatar

@JamesGleick Long ago someone said their weird indenting of autonumbered headings had something intrinsic to do with how they structure docs -- html was involved -- although the point of structured docs is that you can separate formatting from structure. Nevertheless other bugs that have been with the product for decades simply cannot be so intractable. ("simply" = I have no evidence.)

And how does one explain their still producing HTML that uses HTML to mimic UNstructured docs?? aaargh,

dweinberger, to random
@dweinberger@mastodon.social avatar

New research confirms that flipped coins are slightly more likely to land on the side that was up when they were flipped. The difference is less than 1%, so take this as a very long-term betting strategy.

https://phys.org/news/2023-10-flipped-coins-fair-thought.html (via @mathewi)

dweinberger, to random
@dweinberger@mastodon.social avatar

If you wake up without knowing you went to sleep, you may be .

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