ChatGPT achieves a college-level IQ.
OAN’s Roy Francis
10:10 AM – Wednesday, August 2, 2023
A recent study showed that ChatGPT performs at the same level as undergraduate students when answering reasoning questions commonly found on standardized tests.
The University of California Los Angeles conducted a new research study, published in Nature Human Behavior, to assess ChatGPT’s ability to understand and respond to complex information.
According to the study, ChatGPT scored around 80% correct on IQ questions based on Raven’s Progressive Matrices, a nonverbal ability test used for abstract reasoning. This score is higher than the average person’s score of 60% on the same test.
The program also excelled in analogical reasoning, a problem-solving trait previously believed to be unique to humans. It demonstrated the ability to deduce solutions using rational thoughts and logical examples.
Furthermore, ChatGPT scored “better than the average score for humans” when answering analogy questions from the SAT, as reported by the researchers.
While the program’s performance is impressive, concerns have been raised by Furman University assistant philosophy professor Darren Hick. He worries that AI programs like ChatGPT will eventually produce work that is indistinguishable from human work.
Senior study author and UCLA psychology professor Hongjing Lu acknowledged Hick’s concerns, stating, “Surprisingly, not only did GPT-3 perform as well as humans, but it also made similar mistakes. Language learning models are just trying to do word prediction, so we’re surprised they can do reasoning.”
Co-author Keith Holyoak expressed interest in further investigating how ChatGPT learns and receives information, stating, “GPT-3 might be kind of thinking like a human. We’d like to know if it’s really doing it the way people do, or if it’s something brand new — a real artificial intelligence — which would be amazing in its own right.”
This study highlights ChatGPT’s significant progress in natural language understanding, opening up possibilities for its application in customer service, content generation, and academic research.
Although the program continues to improve, it may still provide incorrect or nonsensical answers. OpenAI, the company behind ChatGPT, acknowledges these instances as areas for improvement and is actively working to enhance the model’s capabilities.
It’s important to note that the research primarily used ChatGPT-3, not the newer and more advanced ChatGPT-4 model.
Stay informed! Subscribe to receive breaking news blasts directly to your inbox for free. Subscribe here.
with Eric Scheiner
with Seth Gruber
with Dr. Dave Brat
with Leo Hohmann
By Nette Noestlinger, Matthias Baehr and Elizabeth Howcroft LONDON (Reuters) – Worldcoin will expand its operations to sign up more users globally…
By Munsif Vengattil BENGALURU (Reuters) -Taiwan’s Foxconn will invest $600 million in two projects in Karnataka state in southern India to make…
By Yuvraj Malik (Reuters) -Apple is likely to report a dip in iPhone sales in the April-June quarter as shoppers held out…
By Aditya Kalra, Arpan Chaturvedi and Munsif Vengattil NEW DELHI (Reuters) -Social media platform X is seeking to quash an Indian court…
" Conservative News Daily does not always share or support the views and opinions expressed here; they are just those of the writer."
*As an Amazon Associate I earn from qualifying purchases
Now loading...