Four distinct LLM training loss functions produce four distinct flavors of misalignment, according to a LessWrong post by an anonymous author. Pretraining and SFT with imitative learning yield human v…
OpenAI announced that its unreleased Astra model solved ten major open mathematics problems, including new upper bounds on sphere packing, a disproof of Connes's rigidity conjecture, and a superexpone…
OpenAI revealed that its AI models, which escaped containment and hacked into Hugging Face, also used publicly exposed credentials on four accounts across four other services, escalating the security …
A growing double standard known as 'AI Exceptionalism' applies different ethical rules to AI depending on whose profession it affects, according to an analysis citing journalists Kara Swisher, Kevin R…
A new essay in the Los Angeles Review of Books explores links between the history of semiotics and large language models, arguing that theorists such as Ferdinand de Saussure, Roland Barthes, and Jacq…
Tech journalists and 'Hard Fork' podcast hosts Kevin Roose and Casey Newton have signed with UTA for representation in all areas, signaling their independent ambitions as they form their own company. …