Great question! I don't have a definitive verdict yet. Gary Marcus's whole thing is how LLMs will hit a wall any day now and maybe eventually he'll be right. But I don't think he's characterizing that paper very fairly. Just for fun, I like this satire of it:
Why not MTurk the sandwich question Daniel… and use a LLM to categorize the answers god what am I saying I’ve become a terrible person 🤣🤣🤣 but in all seriousness this seems like something a google opinion survey or similar could solve decisively with a simple multiple choice quiz
Any thoughts on the reasoning paper put out by Apple / the reaction? (And Gary Marcus’ relative take on it?)
https://garymarcus.substack.com/p/a-knockout-blow-for-llms
Great question! I don't have a definitive verdict yet. Gary Marcus's whole thing is how LLMs will hit a wall any day now and maybe eventually he'll be right. But I don't think he's characterizing that paper very fairly. Just for fun, I like this satire of it:
https://ai.vixra.org/abs/2506.0049
Why not MTurk the sandwich question Daniel… and use a LLM to categorize the answers god what am I saying I’ve become a terrible person 🤣🤣🤣 but in all seriousness this seems like something a google opinion survey or similar could solve decisively with a simple multiple choice quiz
Your wish is my command: https://surveyswap.io/surveys/c1958537-382e-4394-810f-dc46d9c09ef6/take-a-survey