The AI-powered chatbot, ChatGBT, has failed miserably in Singapore’s Grade 6 exams, achieving 16% in mathematics and 21% in science.
During the program’s pilot in February, the robot was asked to answer questions from PSLE primary school final exams; These are tests that all 12-year-olds in Singapore must pass, with the aim of determining which secondary schools students will go to after primary school.
PSLE exam questions have been asked Primary on GBT Chat for the years 2020, 2021 and 2022 in Mathematics, Science and English.
The robot achieved an average of 16 out of 100 marks in three math test papers, according to a newspaper The Straits Times.
During the test, the bot could not understand or answer any of the questions about the graphs, and it did not achieve any score on these questions.
GBT Chat also fails in arithmetic operations by adding a number of numbers; When I asked him a simple arithmetic problem, which is summation 60,000، 5,000، 400 And 3was the arithmetic output that guided the robot 65,503However, the correct answer is, according to the Straits Times 65,403.
The AI robot also made serious mistakes in science, achieving an average of 21 out of 100 marks.
Although ChatGPT was able to pass the English language tests, it achieved a lower average, obtaining an average of 11 marks out of 20 in the examination papers submitted to it.
During the English test, the robot encountered a lot of problems, this time with questions and texts containing words with multiple meanings.
An example cited by The Straits Times is the word “value”. or “value”. As he ignored The robot meant the meaning of the word in the context of the text, where the word Value referred to the moral principles of the individual, while his answer revolved around the meaning of the word as if it meant “monetary value”.
But after the robot took the same “repeated tests” last Monday, ChatGPT was able to pass the subjects it failed.
The robot’s inability to pass the sixth-grade exams in Singapore is astonishing – it passed the final exams at the Wharton School of Business, passed exams in four law school courses, and easily passed the medical license exam in the United States.
The New York Times reported that US universities are now revamping testing systems due to concerns that AI bots are being used to cheat. The focus of the exam renewal includes more oral exams, group work and handwritten assessments rather than written online submissions.
ChatGPT was developed by artificial intelligence company OpenAI and launched in November, which claims to have 100 million users by the end of January.
