RT @smllwrlds: 073 of 365 illustrated tiny sci-fi stories: https://t.co/LlsBRRLnLO
@smllwrlds
073 of 365 illustrated tiny sci-fi stories: https://t.co/LlsBRRLnLO
@mrled
Blue skies, Sierra Ferrel, chickenshit bingo. Thanks Luck. https://t.co/kAhLaxX8vm https://t.co/b2O2GfFEeO
@bgolus
Some people say Phillips is good enough, others exclaim the virtues of Torx, and others argue Robinson is vastly superior to all others.
Me personally? I like the friction ball drive type. Guaranteed to never strip! https://t.co/GXSSMRaplC
@ItsSimonTime
Introducing acropalypse: a serious privacy vulnerability in the Google Pixel's inbuilt screenshot editing tool, Markup, enabling partial recovery of the original, unedited image data of a cropped and/or redacted screenshot. Huge thanks to @David3141593 for his help throughout! https://t.co/BXNQomnHbr
@mrled
This would have been my dream laptop until my 2019 onset of RSI :( I miss working out of coffee shops https://t.co/SYAUqXp9Ei
@mrled
TIL: Narrow banking: https://t.co/bckVkBfa6p
@mrled
@joshsojoshsoj To me the question is: if you have trusted inserters, will the public also approve more technical blockchain on top, or is trusting institutions good enough? Some benefits to both, like making fraud harder. Will public understand that tho?
@mrled
@joshsojoshsoj Curious if you have any thoughts about attack surface of the various options. Blockchain proves time of insertion, but you still must trust whatever actors have insert access to the chain. A hybrid solution with only trusted inserters could handle both concerns.
@joshsojoshsoj
It feels fairly obvious that in order for society to continue to function we’re going to need a globally trusted provenance and record keeper
Given the options of governments, companies, and public blockchains, can anyone articulate why a public blockchain isn’t the best choice? https://t.co/uM79sUjeHD
@mrled
PLS DO NOT TAKE https://t.co/g8poxhKKxw
@mrled
The postcards are good, do it https://t.co/yRhWfYN39A
@nickfloats
Midjourney v5 is here! (for real this time, lol)
Here are some side-by-sides of my prompts, v4 vs v5, as well as some new prompts and crowd shots. I'll add more to this as I experiment.
🧵 https://t.co/qSEZWQBXou
@Valdevia_Art
Psst! Hey you! 👀
Want a postcard like these sent to you every month?
Follow me to the reply and join the Patreᴏn! https://t.co/jW31yo5P0h
@smllwrlds
072 of 365 illustrated tiny sci-fi stories: https://t.co/AUwU0TZRW1
@mrled
Like what if all browsers had a subsystem that let this sync to all other browsers you used? https://t.co/PGEKlTWhFP
@mrled
I wish there was like, Fishbowl or Blind but for anonymously sharing outage gossip. Subscribe to channels for the boring enterprise apps you use at work, get notifications that changing your credit card is down because someone UPDATE'd without a WHERE or whatever.
@mrled
Hmm, don't love syndication of longer posts to Twitter. I think I'd rather have a way to edit the tweet text when composing posts, so I could write a summary like "thoughts on localStorage". I've never liked posts that cut off and drop a link at the end. https://t.co/fcOvTEAFr2
@mrled
Imagine browsers offering synced window.localStorage, with an in-browser UI for whether you’re logged in to a site. It would be like iCloud for the web - syncing data via your browser sync password, obviating the need to create a new account for every ... https://t.co/eBUmtcoobU
@mrled
Circa 2000 or so, my family went on vacation to South Carolina to visit my grandparents. Here I am pictured with the most important member of my family (StarCraft) which I brought on the plane and insisted my grandfather install on the PC he used for work. https://t.co/xYes7lCJ7F
@mrled
I have tried to start this experiment several times now, but (as with RPGs) naming is the hardest part. "whispers" isn't that good but maybe it is good enough.
@mrled
Interesting. It doesn't say that it is cross posted, but I made the above post from https://t.co/hP6cyalbqw: https://t.co/Mm7PjvPjtN
Also, apparently it strips out the asterisks I used for bullet points :/
@mrled
Experimenting with a https://t.co/hP6cyalbqw account. Goals:
keep posts on my own domain…
… but also be able to easily write from my phone
write short and medium posts (long ones on main blog)
keep a foot in the door at Twitter by cross posting
@mrled
Birthday party favors :) nonalcoholic Roy Rogers (Coca-Cola and grenadine) on the left and alcoholic painkillers (juice, coconut, rum) on the right https://t.co/h20lGBvbhP
@mrled
Like I'm sorry but this is rude https://t.co/SdbdIzN3jm
@mrled
Well this is the end of that experiment. Tired of seeing this shit 4-5x/day. Fucking disrespectful. https://t.co/MwZKIk9pi2
@mrled
RT @maosboo: Stop looking at your screen—you need to look behind you
> Readers added more context they thought people might want to know…
@mrled
Free white collar crime tip: place criminal communications after the
"
──────────
Hi there,
"
block of a Zoom event and no one will ever find them.
@mrled
When all company events look like this it is Good Actually because now I don't have to be responsible for ever reading a meeting invitation someone sends me https://t.co/vC6UsLhRUP
@mrled
As an: email job haver
I want: all my Google events to contain a list of 50+ Zoom phone numbers
So that: any invitee tempted to start reading my event description will be quickly dissuaded
@maosboo
Stop looking at your screen—you need to look behind you
> Readers added more context they thought people might want to know
The thing standing behind you cannot be seen
> Is this note helpful?
@mrled
@joshsojoshsoj From what I've seen they're at least a bit similar - "walked" is a morpheme "walk" and a morpheme "-ed" for instance. Can get trickier with things like infixes (only present in English with swear words eg absofuckinlute, which is two morphemes)
@joshsojoshsoj
@mrled Oh that’s interesting. I don’t know anything about morphemes. Maybe it’s analogous? That would be interesting to investigate. But yeah if I had to guess, and given the glitches in that computerphile video, they seem very alien
@mrled
@joshsojoshsoj Depending on how you look at it, human language is also tokens, although probably(?) different tokens than ML techniques use. At least in the context of anthropological linguistics, the tokens are usually called morphemes.
@joshsojoshsoj
AI expands the global mind design space, opening the door to minds that are far more alien than any evolution could produce in any physical environment.
Also LLMs aren’t being trained on human language exactly. They’re being trained on tokens https://t.co/P23c0oKccw https://t.co/dPnhE3GFlv https://t.co/AcgzVKsoVv
@mrled
RT @bookazoid_: Yep. We are here. https://t.co/2HYS33d2Sm
@mrled
Are... are those the only two options 😢 https://t.co/IoXbJgCIga
@bookazoid_
Yep. We are here. https://t.co/2HYS33d2Sm
@anthrupad
well that proves it then https://t.co/rxOgUgOAwC
@mrled
This is my biggest concern wrt AI. You can't train them unless you have $billions in hardware and you can't run them without $millions in hardware. It's like running the PC revolution in reverse. https://t.co/8mjXRXILRC
@mrled
I opened a joint checking account in person on Saturday and a personal checking account online this morning. How many fucking checking accounts do u think I need https://t.co/cFIsK8DrvN
@mrled
@VictorLevoso (Interestingly I just reread Revelation Space, which is ~basically about this.)
@mrled
@VictorLevoso The scenarios where like it becomes so malicious/alien that it wants to kill us and can self-improve into a godlike manipulator to get out of its box is hard for me to imagine. If that seemed more plausible then I'd be more worried.
@mrled
@VictorLevoso For me, it is hard to imagine a scenario where something could go off the rails without human intervention. We have non-AI things right now that can kill people; if that happens, we hold the owners responsible. I imagine this is how we'll handle AI-related fatalities too.
@mrled
"microsoft edging"
@mrled
Some users download the Microsoft browser in order to interact with Bing chat, then try to get as close as possible to the line where it refuses to continue a conversation without crossing it. This is a sexual practice known as "Edging".
@VictorLevoso
@mrled Well about that question, how does turning it off when it goes bad look like in your mind?
Like how do we realize the AI is bad before it's too late, given that at this point AI literaly saying it's going to kill everyone wouldn't mean much?
Does it have to kill people?
@balajis
The fact that LLMs are bootstrapped from human language — and not from, say, sensor data — means they arguably are more human-like than a genuine alien.
It’s trained on Reddit posts, not born on a remote planet.