Steven Sinofsky warns anthropomorphizing AI is a critical design error, comparing model disclaimers to ineffective software EULAs

Anthropomorphizing AI will continue to be a design and technology mistake with unbounded implications. The original "I am just a large language model" and subsequent caveats will be the equivalent of EULAs when it comes to understanding limitations.

@krishnanrohit I feel like for most people's purposes here we are talking about 'is there something that it is like to be an LLM' i.e. do LLMs experience quaila

@ishmael5911 Yes. Is there something that it is like to be a major metropolitan US city, is a question you can grammatically ask, without necessarily hitting a useful answer in the binary form.

@krishnanrohit yeah.. Ford Motor Company - feels far closer related to an organism navigation survival through external prompts. no subjective experience (internal prompts). just how i would phrase it. I dont think LLMs can internal prompt as well - they need external prompting to navigate.

@stevesi