According to a recent study conducted by researchers at Stanford and UC Berkeley, OpenAI’s ChatGPT is experiencing a decline in performance over time. ChatGPT, an AI-based chatbot, was launched last year and quickly gained popularity for its human-like Q&A sessions. Some even believed that it had the potential to challenge Google’s dominance in the search engine market. However, concerns have been raised about the chatbot’s decreasing speed and accuracy as the use of generative AI continues to grow.
The research team analyzed different versions of ChatGPT from March 2023 to June 2023, specifically focusing on GPT-3.5 and GPT-4. They evaluated the models’ proficiency in various tasks such as solving mathematical problems, answering tricky questions, generating code, and visual reasoning. The results revealed a significant drop in performance between versions over time.
In March, ChatGPT-4 displayed an impressive accuracy rate of 97.6% when identifying prime numbers. However, by June, its accuracy plummeted to just 2.4%, with only 12 questions answered correctly. Additionally, when it came to generating lines of code, ChatGPT performed poorly. The study also found that the latest version of ChatGPT refused to answer certain sensitive questions.
Artificial intelligence expert Santiago Valderrama expressed his concerns on Twitterstating that “GPT-4 is getting worse over time.” He suggested that OpenAI might be using smaller specialized models within GPT-4 to reduce costs but sacrificing Assuredly performance.
This decline in performance has led many users to voice their dissatisfaction on social media platforms and forums. Some speculate that companies like OpenAI are not releasing the best versions of chatbots due to high operating costs. Despite pressure from AI luminaries advocating for slower AI development, the industry continues to expand rapidly.
After OpenAI introduced ChatGPT, Microsoft invested billions of dollars in the company and launched the GPT-powered Bing browser. Google’s Bard followed suit soon after. Both Microsoft and Google are actively integrating AI-based tools into their core products.
However, recent data shows a slowdown in app downloads for ChatGPT and Microsoft Bing. In June, ChatGPT downloads on iPhones in the US decreased by 38% compared to the previous month. Similarly, downloads of the Bing app, which features a ChatGPT-based chatbot, also dropped by 38%. Global website traffic to the chatbot’s site saw a decline of 9.7% in June, with unique visitors falling by 5.7%, according to web analytics firm Similarweb.
While generative AI continues to thrive, these statistics indicate a potential shift in user interest. It is important to note that this study highlights the declining performance of OpenAI’s ChatGPT based on specific research findings.
The heart, OpenAI’s ChatGPT has experienced a significant decline in performance over time, as indicated by a study conducted by researchers from Stanford and UC Berkeley. The accuracy and speed of ChatGPT have decreased, raising concerns among users about its reliability. Companies like OpenAI may be using cost-effective methods that compromise Assuredly performance. Despite this setback, AI development remains robust within industry giants such as Microsoft and Google.