ChatGPT, the widely used generative AI platform, has sparked concern and confusion among users today as it appears to be generating unexpected and nonsensical responses. OpenAI, the organisation behind ChatGPT, has announced that it is currently investigating reports of these erratic behaviours.
On Thursday morning, OpenAI acknowledged that the chatbot "went a little off the rails yesterday but should be back and operational!"
OpenAI says it found a bug in the part of the system that processes language. "The bug was in the step where the model chooses these numbers" that map tokens that randomly sample words. The company says it has rolled out a fix now and the issue has been resolved.
Social media platforms such as X and Reddit on Wednesday were flooded with complaints from users who encountered unusual responses from ChatGPT. One user on the ChatGPT subreddit expressed bewilderment as ChatGPT suddenly began speaking Spanglish, issuing bizarre threats ("I’m in the room with you right now, lmao"), or simply babbling incoherently. Additionally, when prompted for solutions to math problems or JavaScript bug fixes, ChatGPT delivers paragraphs of random, unrelated material that make no sense in context.
Screenshots shared by users reveal ChatGPT itself acknowledging the issue, stating that a "technical hiccup" caused a nonsensical section to be repeated in its message, and apologizing for the error. Despite numerous reports of similar experiences across social media, it appears that the majority of users remain unaffected by these glitches, with ChatGPT continuing to function normally for many.
While users are sharing a lot of gibberish ChatGPT responses today, this particular example could be fake.
OpenAI has responded to the situation, assuring users that they are actively working to address the issues with ChatGPT and are closely monitoring the situation. While the exact cause of these unexpected responses remains unclear, OpenAI is expected to resolve the matter swiftly to ensure the continued reliability and accuracy of ChatGPT.
As users await further updates and improvements from OpenAI, the incident serves as a reminder of the complexities and potential pitfalls associated with AI technology. While AI platforms like ChatGPT have advanced capabilities in natural language processing and generation, they are not immune to occasional errors or glitches.
Besides, if you are interested in working with ChatGPT makers, then you should know that OpenAI is actively seeking new talent to join its ranks. In a recent tweet, the company's CEO Sam Altman announced that some of the job opportunities are currently available for people to work alongside a highly skilled and collaborative team on the cutting edge of artificial intelligence (AI).
"OpenAI is the most talented and nicest group of people I have ever seen in one place," Altman wrote, further describing the organization as "working on the hardest, most interesting, and most important problems with all the key resources in place."