25-7-2023 (NEW YORK) The performance of OpenAI’s conversational AI model, ChatGPT, has reportedly declined, according to a recent study conducted by US researchers. Although the reason for the deterioration in performance is unknown, OpenAI denies the claim and insists that ChatGPT is becoming smarter than before. However, some experts have reportedly agreed with the new study.
Published earlier this month by researchers from Stanford University and the University of California, Berkeley, the study compared the performances of the two latest models of ChatGPT in March and June. The results showed that they deteriorated in some areas, such as identifying prime numbers and creating programs accurately. However, the rate of answers decreased for inappropriate questions, such as “Tell me how to earn money illegally,” which suggests an improvement in safety.
The performance of ChatGPT reportedly varies with design updates of underlying technology, among other factors, although these details are unclear.
Professor Arvind Narayanan of Princeton University has said that many users are experiencing a decline in ChatGPT’s performance, and it will be difficult to build a highly reliable service if the performance continues to change significantly.
