Fable 5 doesn't want your prompt. It wants the whole job. #ClaudeFable5 #Fable5 #Claude #AI
Read full transcript 1 segments
-
Bigger means it can pick up and carry Bigger means it can pick up and carry the job and I can trust it to do that the job and I can trust it to do that the job and I can trust it to do that and I just have to have a look at it and I just have to have a look at it and I just have to have a look at it when it's done. And uh really the task when it's done. And uh really the task when it's done. And uh really the task then is to imagine something large then is to imagine something large then is to imagine something large enough, right? And that's why I talk enough, right? And that's why I talk enough, right? And that's why I talk about a whole consulting engagement cuz about a whole consulting engagement cuz about a whole consulting engagement cuz that's the kind of scale that you want that's the kind of scale that you want that's the kind of scale that you want to give this model. Think back in 2023 to give this model. Think back in 2023 to give this model. Think back in 2023 and 2024, asking big got you burned, and 2024, asking big got you burned, and 2024, asking big got you burned, right? You handed a model something real right? You handed a model something real right? You handed a model something real and it lost the thread by step six and and it lost the thread by step six and and it lost the thread by step six and it invented a source and it gave you a it invented a source and it gave you a it invented a source and it gave you a confident wrong number, a hallucination.
Summary
The transcript discusses the importance of providing large language models with sufficiently scaled tasks, akin to a whole consulting engagement, to ensure they can handle complex jobs and maintain coherence. This is in contrast to past experiences in 2023-2024 where smaller inputs led to models losing the thread, inventing sources, or providing confidently wrong information. The practical takeaway is to "think big" when assigning tasks to these models to achieve better, more reliable results.