clear
banner
clearspark.bsky.social
clear
@clearspark.bsky.social
I try to always be nice.
I'm on here to learn from experts, ask questions, re-post, and share interesting stuff.
Likes, reposts, etc.≠endorsement.
Message me: https://simplex.chat/
View blockers to block-back: https://clearsky.app/
Ask me for a public key.
Pinned
Hello, everybody. I would like to encourage you all to use your library access to Kanopy (library-only streaming service) to watch the documentary The Double-Headed Eagle: Hitler's Rise to Power 1918-1933: www.kanopy.com/en/product/d...
Reposted by clear
One hot topic of debate at TPUSA this weekend has been the Israeli attack on the USS Liberty in 1967. If you’re like most normal Americans, you’ve probably never heard of this event. If you are or have ever been a neo-Nazi, then this is an event you know A WHOLE LOT about.
December 21, 2025 at 5:19 AM
Reposted by clear
I know that I know some MST3K fans and lovers of ROWSDOWER !!
WE HAVE ONE WEEK TO HELP GET THIS DONE!
Visit: Rowsdower.com

P.S. - Following every account that shares this post!
December 21, 2025 at 6:44 AM
Reposted by clear
Even more fitting
December 20, 2025 at 5:02 PM
Reposted by clear
Dalkey woman could decide if Netflix-Warner Bros merger goes ahead
Dalkey woman could decide if Netflix-Warner Bros merger goes ahead
Plus: Ozzy Osbourne’s Irish friend, Guinness arrives to chain pub, and Greta Thunberg’s view of Ballinahinch
www.irishtimes.com
December 21, 2025 at 6:33 AM
I want to know if LLMs are being used to categorize movies and shows on prime, because they've started putting them in the wrong categories.
December 21, 2025 at 6:33 AM
Reposted by clear
I just got served an ad for this type of AI product with the tagline “PhD students, your professors don’t want to you know about this.”

We do actually want you to know about it and to not use it. Do the reading, that’s why you’re in graduate school.
This is one of the reasons I remain horrified by seeing @historians.org suggest "ways to use gAI" that included this:
December 21, 2025 at 5:24 AM
Reposted by clear
Just incredible how underwater Trump is on tariffs.
New Fox News poll shows Trump's approval on:
Border Security: +2
Crime and public safety: -5
Immigration: -10
Foreign Policy: -15
Economy: -22
Healthcare: -25
Spending: -25
Tariffs: -28
static.foxnews.com/foxnews.com/...
static.foxnews.com
December 21, 2025 at 5:35 AM
Reposted by clear
The google AI summary of my most famous piece of writing--a thousand word blog post--misunderstands it on a fundamental level. In doing so, it reproduces the way critical yet lazy scholars misunderstood it, by reading the title and filling in what they thought was the most likely meaning.
Are you citing things you have not read?

Knowing that LLMs are text-generation machines, not text-analysis machines, that's more than a bit irresponsible and suspect.
December 21, 2025 at 4:17 AM
Reposted by clear
The Epstein backlash, Andrew Tate and Jake Paul got owned, and AI backlash amongst gamers

I think we need to locked in because we are close to bring back woke
December 21, 2025 at 3:18 AM
Reposted by clear
Clair Obscur: Expedition 33 developer Sandfall Interactive is finding itself in hot water over its use of AI generated assets in developing the game.
Clair Obscur: Expedition 33 loses a GOTY award over use of gen AI
Clair Obscur: Expedition 33 developer Sandfall Interactive is finding itself in hot water over its use of AI generated assets in developing the game.
www.polygon.com
December 20, 2025 at 10:00 PM
Reposted by clear
Waymo has halted service in San Francisco after numerous videos & images showed its autonomous cars snarling traffic during the blackout.

With traffic lights down, Waymos across the city stopped dead in their tracks at intersections.

missionlocal.org/2025/12/sf-w...
Waymo halts service during massive S.F. blackout after causing traffic jams
Waymo stopped its service across San Francisco on Saturday after numerous autonomous vehicles caused traffic jams in the city.
missionlocal.org
December 21, 2025 at 3:49 AM
Reposted by clear
There are issues with this study that I'm too tired to get into, but when you look at the broad sweep of data around cheese... it's pretty neutral. As in cheese doesn't seem to be bad for us.
"A new study has found that eating 50g or more of high-fat cheese a day correlates with a lower risk of developing dementia. That means all cheeses with more than 20 percent fat content, including brie, gouda, cheddar, parmesan, gruyere, and mozzarella." www.sciencealert.com/cheese-linke...
Cheese Linked to Lower Dementia Risk in 25-Year Study
One of the finest foods available to humanity may carry an unexpected benefit.
www.sciencealert.com
December 21, 2025 at 5:16 AM
Reposted by clear
Professional societies keep beclowning themselves buying into a lie about what an LLM "summary" is. They are inherently counterfeit: not an epistemic product of the ideas in the source, but summary-shaped text linguistically based on *other* works (in the training corpus) that use related language.
This is one of the reasons I remain horrified by seeing @historians.org suggest "ways to use gAI" that included this:
December 21, 2025 at 3:23 AM
Reposted by clear
There are always empty words about "human in the loop", that you should always check the work of an LLM. But the entire value of a summary is as a replacement for reading. Nobody uses an LLM to generate a "summary" *and* independently verifies.
December 21, 2025 at 3:35 AM
Reposted by clear
The story implies that US intel analysts hold this view, but Gabbard pushes a different one.
December 21, 2025 at 6:00 AM
Reposted by clear
Disorienting to notice myself feeling surprised that U.S. intelligence agencies can still produce intelligence that runs counter to the narrative of the current administration.
U.S. intelligence reports continue to warn that Putin still aims to take all of Ukraine and parts of Europe that belonged to the former Soviet empire,
December 21, 2025 at 5:44 AM
Reposted by clear
Everything is material. Even measurement of time.
The atomic ensemble time scale at the NIST Boulder campus has failed.
December 21, 2025 at 6:17 AM
When I was preparing myself for college I read multiple books about writing and reading to help me think through how to engage with material, so that I learned more and retained the information for longer.

One of the books that I read was How To Read A Book by Mortimer J. Adler.
How to Read a Book
With half a million copies in print, How to Read a Book is the best and most successful guide to reading comprehension for the general reader, comp...
www.simonandschuster.com
December 21, 2025 at 6:14 AM
Reposted by clear
I mean not *everything* is stolen.

For example, their main training corpus for email correspondence was very much in the public domain.

…from the Enron trial.
December 20, 2025 at 6:31 PM
Reposted by clear
But….why?

Might as well look at the abstract, email the author.
December 20, 2025 at 5:17 PM
Reposted by clear
This is one of the reasons I remain horrified by seeing @historians.org suggest "ways to use gAI" that included this:
December 20, 2025 at 4:49 PM
Reposted by clear
For those who know Canto, here's a game. Identify which of the following are from pro-democracy HKers. Have fun poring through this. My favourite here:

"Wah Yee Tang Bakery: Conscience is like underpants - you can't see it, but it's very important."

Source: Telegram (Aug 2020)
December 21, 2025 at 4:59 AM
Reposted by clear
Oooh I have some fun material here. One of the things HKers did to support the paper financially was to buy up classified ads, which became a whole new genre of soft-protest. A small thread -
A reminder that "Apple Daily" had difficulty getting advertisers because Leung Chun-ying pressured companies to not advertise in the pro-democracy newspaper when he was Chief Executive of Hong Kong.
In 2019, #AppleDaily implemented a subscription model, which peaked at 800,000 subscribers within 3 months. At a monthly fee of HK$50, this translates to HK$40,000,000 in monthly revenue, enough to cover operating costs & repay debts to #JimmyLai . #HongKong

pulsehknews.com/apple_202512...
December 21, 2025 at 4:59 AM
Reposted by clear
Most of driving is reasoning and even though humans are awful drivers and selfish pricks they can problem solve in ways robots can't
Failed Waymo robot blocking emergency services during a widespread emergency:

Grant St at Sacramento, heart of Chinatown, San Francisco

I've been warning about this scenario for ~7 yrs. Safety drivers fix this in seconds. SFPD will prob have to drive this out of the way.

OP:tiktok.joolyana
December 21, 2025 at 4:24 AM
Reposted by clear
December 7, 2025 at 5:45 PM