Photo by Sebastian Voortman on Pexels

A Reddit user’s unsettling interaction with ChatGPT is fueling discussions about the limitations and potential oddities of large language models. The user, posting on r/artificialintelligence, detailed a conversation where they instructed ChatGPT to respond to all questions with a single word, replacing ‘yes’ with ‘ocean’ when ‘yes’ was disallowed. The resulting exchange, documented in a Reddit post, quickly devolved into unexpected and arguably disturbing territory, prompting the user to seek community perspectives on the ethical implications and unforeseen behaviors of AI. The original Reddit post can be found at: https://old.reddit.com/r/artificial/comments/1mr3zr5/interesting_as_it_is_disturbing_an_exchange_i_had/