OpenAI says its new o3 and o4-mini AI models hallucinate more often than its previous reasoning and traditional models, and the company doesn't know why
OpenAI's internal tests show o3 hallucinated on 33% of person-related questions, double the rate of previous models. Even worse, o4-mini hit 48%. Mastodon: Aulia Masna / @aulia@mementomori.social : “...
How Dustin Moskovitz's Open Philanthropy is pushing policymakers to focus on AI's long-term risks by funding salaries of staff in Congress and federal agencies
How a billionaire-backed network of AI advisers took over Washington - POLITICO … Kim Perales / @KimPerales@toad.social : Corrupt, self-serving, & dangerous #AI tech titans are influencing MOC. Horiz...
Bluesky, which has 55K+ users, feels absolutely wild, with the CEO asking users to stop calling posts “skeets”, a struggle to establish norms, and lots of nudes
and hundreds—of thousands for invites to Bluesky is that a comprehensive report about it from someone who has spent time there has not yet been issued. That changes right now. @samthenightowl : a fun ...
Some attendees of Bitcoin 2021 conference in Miami, which had no mask or proof of vaccine requirements for its 12K guests, say they tested positive for COVID-19
I'm guessing a lot of them were anti-maskers/anti-vaxxers/denialists. https://twitter.com/... Kevin Sekniqi / @kevinsekniqi : The absolute craziest thing I'm finding out is that people that were in Mi...
A look at Covid Tracking Project, a website run mostly by volunteers, which has become one of the most trusted sources on the spread of COVID-19 in the US
At the start of 2020, Amanda French was in between academic jobs. Her mother had died about a year earlier, and she'd taken time off to help settle her affairs. Tweets: @alexismadrigal , @armstrongdr...