Large language models like ChatGPT are now big enough that they’ve started to display startling, unpredictable behaviors.| Quanta Magazine
According to Stanford researchers, large language models are not greater than the sum of their parts.| Stanford HAI