The information: Meta goes all in on open-source AI. The firm has unveiled LLaMA 2, its first giant language mannequin that’s obtainable for anybody to make use of—at no cost. It’s additionally releasing a model of the AI mannequin that individuals can construct into ChatGPT-style chatbots.
Why it issues: The thought is that by releasing the mannequin into the wild and letting builders and corporations tinker with it, Meta will be taught vital classes about the best way to make its fashions safer, much less biased, and extra environment friendly.
But… Many caveats nonetheless stay. Meta is just not releasing details about the info set that it used to coach LLaMA 2, and it nonetheless spews offensive, dangerous, and in any other case problematic language, identical to rival fashions. Meta additionally can’t assure that it didn’t embody copyrighted works or private information, in response to an organization analysis paper shared solely with MIT Technology Review. Read the total story.
—Melissa Heikkilä
Spotting Chinese state media social accounts continues to be a problem
It’s no secret that Chinese state-owned media are lively on Western social platforms. But typically they take a covert method and distance themselves from China, maybe to achieve extra unsuspecting audiences.
Such operations have been discovered to focus on Chinese- and English-speaking customers up to now. Now, a examine printed final week has found one other community of Twitter accounts that appears to be obscuring its China ties—and this time, it’s made up of Spanish-language information accounts concentrating on Latin America. Read the total story.
—Zeyi Yang