OpenAI confirms ChatGPT getting 'lazier', working on a fix

We’ve been hearing since the last few days that OpenAI’s GPT-4 LLM (Large Language Model) has become lazier,  with users complaining that it does a sloppy job of completing tasks it used to do with ease sometime back. It’s now been confirmed that those complaints were indeed genuine, as OpenAI has recently acknowledged that GPT-4 has become ‘lazier’.

In a post on X (formerly known as Twitter), OpenAI explained why the GPT-4 has become what it’s become. The reason is that the AI tech firm hasn’t updated the model since November 11, which is exactly a month ago. The company hasn’t given any details on why it hadn’t updated the model, but it does say that the move wasn’t “intentional”. 

OpenAI is working on a fix to this issue, but when it’ll be available for users isn’t known yet. In the meanwhile, if you’re using ChatGPT, which is powered by GPT-4, you might see performance degrade only in a subset of prompts. Again, OpenAI hasn’t disclosed the prompts that the LLM does a sloppy job of answering. 

As a reminder, GPT-4 is the backbone of many AI models other than ChatGPT. However, it’s not exactly clear how not updating GPT-4 for a month impacts the performance of other AI models like Microsoft’s Copilot. If you’re not happy with the performance of GPT-4, you can start using Google’s recently launched Gemini, because the latter is as good as the former. 

Have you experienced that GPT-4 has become ‘lazier’? You can share your experience with us in the comments section below.