drsezzer.bsky.social - Profile | ThreadSky | a Reddit-style client for Bluesky

Saturday morning mood! What's Up by 4 Non Blondes And so I wake in the morning And I step outside And I take a deep breath and I get real high And I scream from the top of my lungs What's going on? #MusicSky #Revolution youtu.be/6NXnxTNIWkc?...

submitted 11 hours ago • 22 comments

💻 A lot of ethically problematic details in Anthropic's new marketing of its Claude model. 🧵 1. The use of the term "whistleblow" is interesting. IANAL, but whistleblowing would require continual surveillance of system-user interactions that are exfiltrated without user knowledge or consent.

submitted 1 day ago • 3 comments

So Claude blackmails people and OpenAI is building a 24/7 surveillance tool. Black Mirror was a warmup act.

submitted 2 days ago • 3 comments

Wow... Comment in thread about Anthropic's idea of the model contacting authorities or media if user is trying to do something it determines is not allowed. So many #facepalms

submitted 1 day ago • 0 comments

MOON'S WINDOW (2007) Acrylic on Panel - 7" X 5" This window was suggested by shapes at the Whitby Abbey, Yorkshire England. (Nothing to do with the painting, but Whitby is where Count Dracula landed when he arrived in England...) 1/3

submitted 2 days ago • 9 comments

ChatGPT's new dossier-from-your-chats feature is a huge change to how it works, and as a power user who tries to control all of the model's input I don't like it at all “30 messages are good interaction quality (25%); 9 messages are bad interaction quality (7%)” simonwillison.net/2025/May/21/...

submitted 3 days ago • 14 comments

Doug Cockle is the best, and he's not wrong. Read the sodding books! The Witcher has always been about Ciri.

submitted 3 days ago • 0 comments

Oof. on.ft.com/45k3Dcg M&S blames ‘human error’ for cyber attack that will hit profit by £300mn

submitted 3 days ago • 3 comments

🚨New WP🚨 Can AI do "deep canvassing"—the time-intensive, empathic persuasive dialogues that durably reduce prejudice when done by humans? Spoiler: Yes! We find durable reduction in prejudice toward undocumented immigrants LLMs make deep canvassing possible at massive scale osf.io/preprints/os...

submitted 3 days ago • 3 comments

Nope nope nope

submitted 3 days ago • 1 comment

"LLMs do not distinguish between fact & fiction. They are not designed to tell the truth. They are designed to persuade, yet they are implemented in sectors where truth & detail matter, e.g. education, science, health, the media, law, & finance." My interview @washingtonpost.com tinyurl.com/2e7253ja

submitted 3 days ago • 6 comments

VEO3 Prompt: “a man doing stand up comedy in a small venue tells a joke (include the joke in the dialogue)” The video and audio are generated together.

submitted 4 days ago • 180 comments

"I want people to be safe and protected. I want to live on the right side of history. I am an immigrant. My parents are refugees from Chile. We fled a dictatorship and I was privileged enough to grow up in the United States after asylum in Denmark. I stand by those protections, always."

submitted 6 days ago • 317 comments

Took Lugnut out on a play date. With fellow robot maker davis #robot

submitted 7 days ago • 3 comments

This is a good experiment, but important caveat on p. 18. The effect is not because LLMs have some fiendishly clever manipulation strategy. It's mainly that they're fast, fluent, and patient — and these are effects that wear off a bit as people get used to them.

submitted 7 days ago • 0 comments

Unexpected things you hear when teenagers mutter as they leave their rooms for snacks... #48: "I should have mentioned Thomas Aquinas, FFS".

submitted 7 days ago • 1 comment

🎶 Imagine 🎼 🎶🎶

submitted 7 days ago • 556 comments

Phoenix Doom, a PC port of the 3DO version of DOOM with several quality of life improvements: github.com/BodbDearg/ph...

submitted 9 days ago • 2 comments

Announcing AlphaEvolve, our new LLM coding agent that has - made new scientific discoveries - discovered algorithms that are now deployed at Google (in Gemini, Transformers, TPU hardware design & data centers) Blog: deepmind.google/discover/blo... White paper: storage.googleapis.com/deepmind-med...

submitted 10 days ago • 6 comments

'this new species of agents...' how do they get away with this type of terminology? #facepalm

submitted 10 days ago • 0 comments

*crab-walked

submitted 10 days ago • 408 comments

"Women are PIs on 58% of the canceled grants, although they are PIs on only 34% of all active NSF grants. Similarly, Blacks are PIs on 17% of the terminated grants, although they make only 4% of the total pool. Hispanic PIs and those with disabilities were twice as likely to lose a grant."

submitted 11 days ago • 32 comments

One of the greatest tech tragedies of our time is the way we’ve let the phone and post office degrade into a basket of scams, exposing millions of people to risk on a daily basis.

submitted 11 days ago • 17 comments

In huge news, the UK government was just defeated in the House of Lords for a second time over AI & copyright. The government had removed amendments to the Data Bill that would require AI companies to disclose their training data. The Lords have now reinserted them. 🧵 1/2

submitted 12 days ago • 9 comments

Character dot AI is trying to get out of a wrongful death lawsuit by claiming that AI has First Amendment protection of its speech mashable.com/article/chat... It should be needless to say, but computer software are not persons under the law.

submitted 12 days ago • 2 comments

So frustrating watching AI types relearning what true software engineering and basic IT security is, all over again!

submitted 12 days ago • 0 comments

As AI-driven fraud becomes increasingly common, more people feel the need to verify every interaction they have online.

submitted 12 days ago • 7 comments

Marking UG/MSc projects in applied AI. Cant discuss individual cases, but some students spend most of their time understand domain, user requirements, cleaning data; and others spend almost all their time building and testing models. First group is closer to my perspective!

submitted 12 days ago • 0 comments

For everyone who was like “Google search is getting worse,” we were absolutely right.

submitted 12 days ago • 319 comments

This will rankle those that gatekeep science through the use of inaccessible language! Waiting for the push back...🍿

submitted 12 days ago • 0 comments

I figured out that misogyny was bad before I even had a daughter. But thats only because im a genius

submitted 14 days ago • 7 comments

Knowing how to use a tool. Is part of getting the best results. Knowing you're dealing with something that might know, but might make it up... should hone critical thinking skills (as long as you stay cognizant of the fact over time).

submitted 13 days ago • 0 comments

The site that gives the wrong stats is Barrat homes (builders in UK), the page seems to confuse number of Google searches for charity shops per capital with .. number of shops per capita. I suspect the reason it's returned this page is (I've discounted paid advertising). It's a sensational number!

submitted 13 days ago • 1 comment

Holy shiitake! Trump fired Shira Perlmutter, the Director of the US Copyrights Office apparently because she refused to let Elon Musk raid the place to steal author’s intellectual property rights democrats-cha.house.gov/media/press-...

submitted 13 days ago • 137 comments

Cream-spotted Tigerwing from this morning at @tapirvalley.bsky.social #butterflies #costarica #nature 🌿

submitted 15 days ago • 24 comments