Conversation

Notices

1. Embed this notice
   Eaton (eaton@phire.place)'s status on Friday, 24-May-2024 12:13:04 JST Eaton

   in reply to

   • Dare Obasanjo

   @carnage4life I’ve been saying for a while that without realizing it we collectively overindexed on “conversational english” as the signal of AI advancement; LLMs absolutely smoke that test, but reveal a huge iceberg of meaning/accuracy/model-of-the-world problems underneath. It’s a pivot to video level mistake at the industry level, IMO — not an inherently bad tech/medium/tool, but a huge expensive waste when splattered onto everything.

   • Embed this notice
     Dare Obasanjo (carnage4life@mas.to)'s status on Friday, 24-May-2024 12:13:05 JST Dare Obasanjo

     I sometimes wonder if LLMs are this generation’s self driving cars? A decade ago self driving cars were so promising it was assumed they were 90% of the way to being ubiquitous. But now it looks like the last 10% of the work is taking 90% of the time.

     LLMs work well 90% of the time. The question is how hard will it be to address that last 10%?