François Chollet on Twitter: "While large language models like GPT-3 can't do most of what humans do, let's remember that the reverse is also true: humans aren't trained to complete text from arbitrary prompts, and would do a terrible job at it for most prompts..."

Country	Code	For customers of
United States	40404	(any)
Canada	21212	(any)
United Kingdom	86444	Vodafone, Orange, 3, O2
Brazil	40404	Nextel, TIM
Haiti	40404	Digicel, Voila
Ireland	51210	Vodafone, O2
India	53000	Bharti Airtel, Videocon, Reliance
Indonesia	89887	AXIS, 3, Telkomsel, Indosat, XL Axiata
Italy	4880804	Wind
3424486444	Vodafone
» See SMS short codes for other countries

1. François Chollet‏Verified account @fchollet 1 Sep 2020
  A language model is just very, very different from a mind, and any overlap in capabilities is an accidental artifact.
  
  7 replies 45 retweets 422 likes
  
  Show this thread
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Setareh‏ @starrezaee 1 Sep 2020
  Replying to @fchollet
  
  Always I ask myself, why people think the AI or robotics might be look like humans?
  
  1 like
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Ruchit G Garg‏Verified account @ruchitgarg 1 Sep 2020
  Replying to @fchollet
  
  Maybe we are specialized robots too..
  
  1 retweet
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Stefan Güttel‏ @guettel 1 Sep 2020
  Replying to @fchollet
  
  Yes, but humans do things intentionally, with some kind of objective in mind. Completing a text randomly using a large corpus of human-written text is merely a compilation task. Humans are "terrible" at it because it's an ill-posed objective.
  
  5 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. ‏ @thecarrozo 1 Sep 2020
  Replying to @fchollet
  
  Ha! I’m attempting to do precisely this with "GPT-36” :) Doing a max of 36 of these, deadline to submit this Thursday: https://carrozo.substack.com/p/exclusive-access-to-gpt-36 …
  
  3 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Kishore Seshagiri‏ @seshak 1 Sep 2020
  Replying to @fchollet
  
  How do you see the evolution of NLP with machines starting to understand language better and humans starting to lose language comprehension with digital era taking over
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo