This is an important problem, as we've mentioned, of close nation and this appears to be a pretty good breakthrough on that. They found that using their technique, certain kinds of falsehoods can get weeded out of the system more easily than others. It's kind of interesting to see that like the model seems to understand truthfulness as a concept for some fields, but then for other fields, it seems to be maybe a little less good at internally recognizing the truth value of these statements. And one last thought here is they do build on top of llama.

Get the Snipd
podcast app

Unlock the knowledge in podcasts with the podcast player of the future.
App store bannerPlay store banner

AI-powered
podcast player

Listen to all your favourite podcasts with AI-powered features

Discover
highlights

Listen to the best highlights from the podcasts you love and dive into the full episode

Save any
moment

Hear something you like? Tap your headphones to save it with AI-generated key takeaways

Share
& Export

Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more

AI-powered
podcast player

Listen to all your favourite podcasts with AI-powered features

Discover
highlights

Listen to the best highlights from the podcasts you love and dive into the full episode