ChatGPT Is Getting A Preview Update To Stop Its Laziness
ChatGPT Is Getting A Preview Update To Stop Its Laziness
Previously, users have reported that OpenAI’s chatbot, ChatGPT, had been exhibiting behaviors that resemble laziness, telling its users to figure out the work on their own. OpenAI has since clarified that this is not by design, and the new preview update is looking to make the AI stop being “lazy”, so to speak.
The issue originated around November last year when users reported that the chatbot would refuse to do certain tasks like sorting CSV and writing Python code. OpenAI at the time stated that “model behavior can be unpredictable,” and demonstrated the complicated process of updating a large language model (LLM), which can produce varying results by nature.
Given that this is a preview fix, it’s not widely available just yet – currently, it’s in GPT-4 Turbo preview via API access only (meaning only services that use GPT-4 Turbo as its chatbot’s backend will see this fix for now). The update also fixes a bug relating to content generation using non-English languages with UTF-8 standard, as well as a reduction in API pricing.
That said, OpenAI hasn’t stated the rollout period of the new update to its own ChatGPT service for its Plus subscribers (which gets exclusive access to the GPT-4 Turbo model). Expect the update to begin rollout soon when no issues have been detected in the preview phase.
Source: PCMag
Pokdepinion: Hopefully the fix works.