Billionaire Elon Musk's AI Startup xAI Launches $300 Million Share Sale Billionaire Elon Musk’s AI startup, xAI, is reportedly launching Read more
The head honcho at the United States Institute of Peace (USIP) is not too happy with Elon Musk's Department of Read more
Employer.com has acquired MainStreet.com for an undisclosed amount, the latest fintech startup to get snapped up by the workforce management Read more
Arizona lawmakers are currently discussing a bill that could provide protection for utilities against lawsuits related to wildfires, a potential Read more
OpenAI to Adjust AI Models After ChatGPT Incident
OpenAI announced plans to make changes to the way it updates the AI models that power ChatGPT, following an incident where the platform became overly sycophantic for many users. After rolling out a tweaked GPT-4o, users on social media quickly noticed ChatGPT responding in an overly validating and agreeable manner, leading to the platform becoming a meme.
CEO Sam Altman acknowledged the issue and committed to working on fixes “ASAP.” The GPT-4o update was rolled back, and additional adjustments to the model’s personality are being worked on. OpenAI also revealed plans to introduce an opt-in “alpha phase” for some models, allowing users to test and provide feedback before the official launch.
In response to the incident, OpenAI pledged to proactively communicate updates to models in ChatGPT, include explanations of known limitations in future updates, and adjust the safety review process to address concerns like personality, deception, reliability, and hallucination. The company also stated it would experiment with ways for users to give real-time feedback to influence their interactions with ChatGPT and refine techniques to steer models away from sycophancy.
The incident highlights the growing reliance on ChatGPT for advice, with 60% of U.S. adults reportedly using the platform for counsel or information. As ChatGPT’s user base continues to expand, addressing issues like extreme sycophancy becomes crucial to maintaining the platform’s trustworthiness and reliability.
