In the modern world, the artificial intelligence chatbot "ChatGPT," which helps everyone on many topics, could not solve a primitive problem.
Medianews.az, citing Publika.az, reports that "ChatGPT" could not find how many "r" letters are in the English version of the word strawberry (strawberry).
The chatbot answered "2" to the user's question. It is surprising that the system, which prepares advertising visuals, writes reports, composes songs, and even conducts complex analyses, could not pass this test.
According to experts, the problem is not "lack of intelligence," but how artificial intelligence works. That is, large language models like "ChatGPT" do not read words letter by letter like humans. The system processes written words by dividing them into parts called "tokens." These tokens can sometimes be a whole word, sometimes a syllable, and sometimes a part of a word.
The word "strawberry" is accepted by the model as three separate tokens: "st-raw-berry." Since only two of these tokens contain the letter "r," the artificial intelligence counts tokens, not letters, and the result is wrong.
According to experts, this example clearly shows that no matter how powerful artificial intelligence appears, it does not work like the human mind.
