OpenAI has recently announced enhancements to its chatbot, ChatGPT, which now correctly identifies the number of occurrences of the letter “R” in the word “strawberry.” This improvement comes after users frequently encountered inaccuracies, with the chatbot often asserting that there are not three “R”s in the word. Despite this update, other issues persist, as users continue to report incorrect responses on similar prompts.
The company shared its achievement on Twitter/X, celebrating the milestone as a sign of progress in AI development. However, many responses to the post highlighted that ChatGPT still struggles with basic logic, such as advising users to walk to a car wash only 50 meters away instead of driving. This inconsistency raises questions about the reliability of AI models.
OpenAI’s adjustments seem to be hardcoded solutions rather than comprehensive fixes for underlying problems. For instance, queries about the number of “R”s in other words like “cranberry” still yield incorrect answers, further indicating that while improvements have been made, challenges remain in ensuring accuracy and logical reasoning in AI responses.