Last year, the company Open AI launched the chatbot ChatGPT, which can, among other things, answer questions, formulate emails and write speeches for you.
Recently, ChatGPT has also been tested on several demanding exams – where it has passed.
It writes Business Insider.
Law and medicine
Professor Christian Terwiesch, at Wharton University of Pennsylvania, recently tested the AI robot on an operations management exam.
According to Terwiesch, the technology was able to answer basic questions based on case studies, and the robot did a “fantastic job”.
It made some mistakes on simple math problems, and some more demanding questions, but would still get a B or B- on the exam.
according to CNN If the AI robot also passed law exams in four courses at the University of Minnesota Law School, it would have received a grade of C+. Here, the robot answered multiple-choice questions and longer text tasks.
In addition to these, ChatGPT has also passed exams in, among other things, medicine at the United States Medical Licensing Exam and biomedicine at the School of Health Sciences at Høyskolen Kristiania.
Vice-dean for education at the School of Health Sciences, Hilde Skjerve, has previously told the online newspaper Chrono that the AI robot received grades from B to E in the various exam parts for biomedicine.
The survey also showed that the answers from the robot became worse when the questions on the exam became more complex.
AI on home exam
AI technology has gone viral recently, and there are divided opinions on whether the function is positive or negative.
Especially when it comes to cheating on school exams.
In December, Bloomberg podcaster Matthew S. Schwartz signed on, among others Twitter that “home exams are dead”.
He had tested ChatGPT on a legal assignment where he claimed that the AI technology gave a quick and solid answer.
Previously, TV 2 wrote that almost no universities in Norway knew how to reveal AI use in exams where all aids were allowed before Christmas.
In January, only three Norwegian universities and colleges reported that they had discovered use of the robot in connection with examinations.
The Norwegian Directorate of Education has also chosen to limit internet access for some upper secondary school exams in order to prevent the use of ChatGPT.
Almost as quickly as ChatGPT became a talking point, the news also came about Edward Tian (22) who has created a robot that detects the use of ChatGPT.
Tian is a student at Princeton University and spent New Year’s Eve developing the GPTZero application.
The application works in such a way that if GPTZero is confused by a text, there is a high probability that the text was written by a human. If the application is not confused, the text is probably written by an AI robot.
I spent New Years building GPTZero — an app that can quickly and efficiently detect whether an essay is ChatGPT or human written
— Edward Tian (@edward_the6) January 3, 2023
