NEW: DHS's "chief AI officer" and head of its biometric surveillance operation came to the agency straight from defense tech vendor Anduril — a position conveniently kept off his public resume. www.levernews.com/the-spy-tech...
”they could start by providing more internet. We get a data allowance for free, and then we need to pay…but the internet isn’t always available because the GPS signal here gets jammed by the Iranians or the Americans” www.theguardian.com/business/202...
”they could start by providing more internet. We get a data allowance for free, and then we need to pay…but the internet isn’t always available because the GPS signal here gets jammed by the Iranians or the Americans” www.theguardian.com/business/202...
Is this you
this creepy bible thumper is going to be a major player in UK politics by the end of the decade
www.theguardian.com/music/2026/m...
unless i’m missing it here…they’re not even bothering to list the CPU clock speed on the specs page!?
www.apple.com/ca/macbook-n...
unless i’m missing it here…they’re not even bothering to list the CPU clock speed on the specs page!?
www.apple.com/ca/macbook-n...
Introducing the Macbook Neo™
whoa! amazing! thank you, hope it was useful!
Today my AI in Fact & Fiction class is discussing 2 short stories: Ray Bradbury's _The Veldt_ and @timmaughan.bsky.social's _Be Good for Goodness' Sake_. We're talking broadly about surveillance capitalism in relation to where all the fucking data comes from for what gets called AI now.
To lendo um livro de sci-fi, Infinite Detail do Tim Maughan; rola um ataque cibernético q acaba c/ a internet globalmente num estágio do capitalismo de vigilância em q as bigtechs dominaram td q era público, e retrata o colapso da sociedade. Penso nesse livro sempre q vejo notícias do tipo.
very briefly flashed up at start of a youtube video: Disclaimer In this video, l've highlighted some of the best GNOME extensions. While extensions are usually safe to use, please be mindful that they might cause system instability or pose privacy risks. Also, this video uses an Al-generated voiceover, which may not sound fully natural at times.
oh shit your video sounds great man yeah i’ll be sure to both like and subscribe
is it? i’d been kinda almost actively avoiding it lol. by which i mean youtube really wants me to watch it but every time i’m like ehhhh maybe later
one thing we maybe don’t talk enough about “these days” is how you encounter people that will just straight up lie to you - IRL, to your face - about things you both know aren’t true
(btw this is also why polling doesn’t work and you should never trust any of it)
*it’s also on spotify apparently, but i’d rather you stole it
*it’s also on spotify apparently, but i’d rather you stole it
the audiobook of #infinitedetail is available at all good stores and torrent sites www.audible.ca/pd/Infinite-...
pretty on point Chris Tapsell write up www.eurogamer.net/future-of-xb...
Link from Legend of Zelda pointing to an open grave with his thumb. A word bubble caption shows him saying "Get in"
Microsoft to Xbox:
classic jeff
"What's that Judge Tony? Tough on crime, tough on the causes of crime?"
also i just don't believe them - the 'open data' they're boasting about includes "1 million real-world user-ChatGPT interactions" and "diverse mix of web content, academic publications, code, books, and encyclopedic materials"
it's the exact same shit
allenai.org/open-data
any locally run LLM - regardless of what datasets you plan to feed it - has already been trained on personal data/stolen work, on thousands of GPUs, in a huge power/water hungry data center, for months
its why you should laugh at any "artist" that claims they only trained an LLM on their "own work"
*taps sign* there's no such thing as a small language model
eh, one person's "important distinction" is someone else's whataboutism i guess
tbh it's only an important distinction if you really believe that resources are the only pressing ethical problem that LLMs present
eh, one person's "important distinction" is someone else's whataboutism i guess
tbh it's only an important distinction if you really believe that resources are the only pressing ethical problem that LLMs present
Similar to how Grokipedia took a snapshot of Wikipedia, used AI to remove “woke stuff” but now is becoming rapidly out of date on current events and celebrity deaths because it doesn’t have an army of people updating it.
any locally run LLM - regardless of what datasets you plan to feed it - has already been trained on personal data/stolen work, on thousands of GPUs, in a huge power/water hungry data center, for months
its why you should laugh at any "artist" that claims they only trained an LLM on their "own work"
the "open source LLMs running locally" idea is very naive
nobody in the OS community has the resources to build an LLM "from scratch" - mainly ppl use a variant of Llama - meta's LLM - trained on stolen data that was cleaned by underpaid workers in global south - and thus exhibit the same biases
